Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

REVIEW

Emerging insights into the molecular and cellular basis of glioblastoma

Gavin P. Dunn,1,2,3 Mikael L. Rinne,2,3,4 Jill Wykosky,5 Giannicola Genovese,2 Steven N. Quayle,2 Ian F. Dunn,6 Pankaj K. Agarwalla,1,2,3 Milan G. Chheda,2,3,7 Benito Campos,8 Alan Wang,9 Cameron Brennan,10 Keith L. Ligon,11 Frank Furnari,5,12 Webster K. Cavenee,5 Ronald A. Depinho,9 Lynda Chin,13 and William C. Hahn2,3,14 1Department of Neurosurgery, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts 02114, USA; 2Department of Medical Oncology, Division of Molecular and Cellular Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts 02115, USA; 3Broad Institute of Harvard and MIT, Cambridge, Massachusetts 02142, USA; 4Center for Neuro-Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts 02215, USA; 5Ludwig Institute for Cancer Research, University of California at San Diego School of Medicine, La Jolla, California 92093-0660, USA; 6Department of Neurosurgery, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts 02115, USA; 7Department of Neurology, the Brain Tumor Center, Memorial Sloan Kettering Cancer Center, New York, New York 10065, USA; 8Department of Neurosurgery, University of Heidelberg, Heidelberg, Germany; 9Department of Cancer Biology, The University of Texas MD Anderson Cancer Center, Houston, Texas 77030, USA; 10Department of Neurosurgery, the Brain Tumor Center, Memorial Sloan Kettering Cancer Center, New York, New York 10065, USA; 11Department of Medical Oncology, Center for Molecular Oncologic Pathology, Dana-Farber Cancer Institute, Boston, Massachusetts 02215, USA; 12Department of Pathology, University of California at San Diego, La Jolla, California 92093-0660, USA; 13Department of Genomic Medicine, Division of Cancer Medicine, The University of Texas MD Anderson Cancer Center, Houston, Texas 77030, USA

Glioblastoma is both the most common and lethal pri- and temozolomide therapy. Unfortunately, current stan- mary malignant brain tumor. Extensive multiplatform dard-of-care therapy results in a median survival of only genomic characterization has provided a higher-resolu- 12–15 mo (Stupp et al. 2005). tion picture of the molecular alterations underlying this The emergence of a molecularly focused approach to disease. These studies provide the emerging view that cancers represents a profound shift in our approach to the ‘‘glioblastoma’’ represents several histologically similar diagnosis and treatment of malignancy. This framework yet molecularly heterogeneous diseases, which influ- centers on a new taxonomy that describes cancers by ences taxonomic classification systems, prognosis, and their molecular alterations and the identification of in- therapeutic decisions. hibitors that target these cancer-specific changes. In- deed, therapies targeting HER2-amplified (Slamon et al. 2001), CML harboring the BCR–ABL In his effort in the 1920s to cure two comatose patients translocation (Druker et al. 2001), mutant EGFR lung with extensive glioblastomas, neurosurgeon Walter Dandy took the radical step of removing the entire affected cancer (Lynch et al. 2004), lung cancer harboring the hemisphere in each of these patients. Despite these heroic EML4–ALK translocation (Kwak et al. 2010), and BRAF interventions, both of these patients succumbed to their mutant melanoma (Chapman et al. 2011) provide clear disease due to involvement of the contralateral hemisphere proof of principle for this approach. Histopathologic (Dandy 1928). More than 80 years later, we continue to face diagnosis is increasingly being supplemented with annota- the same clinical challenges illustrated by this vignette tion of amplification or mutational status where relevant from the early 20th Century. Glioblastoma represents the (MacConaill et al. 2009; Dias-Santagata et al. 2010), and most common primary intrinsic malignant brain tumor these data subsequently inform therapeutic decision-mak- diagnosed each year in the United States; there are ;10,000 ing, all in a time frame from target discovery to therapy new diagnoses annually, and >50,000 patients are currently that is becoming progressively shorter (Chabner 2011). living with the disease (Davis et al. 2001; Porter et al. 2010; However, despite clinical progress across many cancer CBTRUS 2011). The clinical hallmarks of glioblastoma are types and extensive characterization of genomic alter- its aggressive growth and inexorable recurrence despite ations in glioblastoma, we still have not identified and multimodal therapy with surgery followed by radiation exploited clinically meaningful tumor dependencies in this dreaded disease. [Keywords: glioblastoma; invasion; angiogenesis; tumor-initiating cells; The recent characterization of the genome (Beroukhim IDH1; glioma classification] et al. 2007; The Cancer Genome Atlas Research Network 14Corresponding author. E-mail [email protected]. 2008; Parsons et al. 2008) and transcriptome (Phillips et al. Article is online at http://www.genesdev.org/cgi/doi/10.1101/gad.187922.112. 2006; Verhaak et al. 2010) of glioblastoma provides a high-

756 & DEVELOPMENT 26:756–784 Ó 2012 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/12; www.genesdev.org Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma resolution picture of the glioblastoma landscape that has as well as focal events involving one or revealed the major structural and expression alterations a few genes. The development of high-resolution, array- that may drive disease pathogenesis and biology. These based comparative genomic hybridization (CGH) using comprehensive data sets reveal ‘‘glioblastoma’’ as a het- bacterial artificial chromosomes, oligomers, and single- erogeneous collection of distinct diseases with multiple nucleotide polymorphism (SNP) arrays made possible dependencies both within and across each particular sub- the systematic analysis of cancer genomes and defined type. Here we summarize recent efforts to catalog the many new recurrent SCNAs in cancer (Chin et al. 2011). structural genomic landscape of glioblastoma and focus on To fully explore these data, several groups have also emerging insights into critical molecular pathways cen- developed bioinformatic tools to highlight the cancer- tral to glioblastoma pathobiology. derived genomic alteration signals from among the back- ground noise (Beroukhim et al. 2007; Greenman et al. 2007; Wiedemeyer et al. 2008; Riddick and Fine 2011). Current histopathological classification These methodologies, including Genomic Identification of Significant Targets in Cancer (GISTIC) (Beroukhim et al. As described by the World Health Organization (WHO) 2007; Mermel et al. 2011) and Genomic Topography Scan classification (Louis et al. 2007), malignant diffuse glio- (GTS) (Wiedemeyer et al. 2008), were first deployed in mas are comprised of astrocytic, oligodengroglial, and glioblastoma and subsequently across thousands of cancer mixed oligoastrocytic neoplasms based solely on mor- samples (Beroukhim et al. 2010; http://www.broadinstitute. phology and are further subdivided by tumor grade based org/tumorscape). Combined work using these tools and on additional histologic features present in the tumor. others (Bredel et al. 2005; Kotliarov et al. 2006) identified Nuclear atypia and mitotic activity are required criteria overlapping presumed targets of amplification (including for grade III lesions, and the presence of necrosis or EGFR, MET, PDGFRA, MDM4, MDM2, CCND2, PIK3CA, microvascular proliferation is required for the diagnosis MYC, CDK4,andCDK6) and deletion (including CDKN2A/ of grade IV astrocytoma, glioblastoma (Miller and Perry B, CDKN2C, PTEN,andRB1) in glioblastoma. 2007). While 90%–95% of glioblastomas arise de novo This work in SCNAs preceded two recent studies that and are considered ‘‘primary,’’ ;5%–10% arise from lower- further clarified our understanding of the glioblastoma grade gliomas in younger patients and are termed ‘‘sec- genomic landscape (The Cancer Genome Atlas Research ondary’’ (Biernat et al. 2004; Ohgaki and Kleihues 2005). Network 2008; Parsons et al. 2008). The Cancer Genome Together with grade III anaplastic astrocytoma, which has Atlas (TCGA) pilot project, the first of many ongoing an incidence of >1500 cases annually (CBTRUS 2011), comprehensive TCGA consortium-based cancer genome these tumors comprise the clinical entity termed ‘‘ma- analyses (http://cancergenome.nih.gov), applied multiplat- lignant glioma.’’ However, as discussed below, emerging form profiling to systematically and comprehensively define classification schemes are likely to incorporate molecular the genomic landscape of glioblastoma. This approach used differences that distinguish these tumors (von Deimling targeted Sanger sequencing to investigate 601 genes in 91 et al. 2011). samples as well as array-based platforms to analyze copy number, mRNA expression, and the epigenetic state of Large-scale genomic studies of glial neoplasms ;200 tumors, most of which were untreated, primary glioblastomas. This larger effort confirmed prior SCNA Glioblastoma, like other cancers, is the product of accu- analyses on smaller numbers of tumors and also identi- mulated genetic and epigenetic alterations (Vogelstein fied novel lesions, such as homozygous deletions of PARK2 and Kinzler 2004; Weir et al. 2004; Stratton et al. 2009), and amplification of AKT3. Moreover, separate work also and the application of genome-scale approaches to enu- showed that NF-kBinhibitora (NFKBIA) is heterozygously merate these genetic alterations has uncovered both deleted in a mutually exclusive manner to EGFR amplifi- molecular subclasses and common pathways mutated cation (Bredel et al. 2011). in this disease. To date, glioblastoma has been subjected The most significantly somatically mutated genes (false to the most extensive genomic profiling of any cancer, and discovery rate <0.1) among the 601 genes analyzed were thus we review here our current understanding of the TP53 (42%), PTEN (33%), neurofibromatosis-1 (NF1) critical underlying molecular pathology in this disease that (21%), EGFR (18%), RB1 (11%), PIK3R1 (10%), and has come from an integrated view of somatic copy number PIK3CA (7%). However, this effort demonstrated that alterations (SCNAs) and sequence mutations and also the integration of multiple types of genomic analysis review recent discoveries in lower-grade astrocytomas. permitted the projection of identified alterations onto known pathways (Fig. 1). This approach revealed the high incidence of , Rb, and tyrosine Integrated approaches to characterize the genomic (RTK)/Ras/phosphoinositide 3-kinase (PI3K) pathway landscape of glioblastoma dysregulation, confirming previous work that had de- Malignant gliomas, like other human cancers, are char- lineated lesions in these critical cascades (Ekstrand et al. acterized by genetic instability and complex alterations 1991; Henson et al. 1994; Louis 1994, 2006; Reifenberger in structure and copy number. The SCNAs et al. 1994; Schmidt et al. 1994; Ueki et al. 1996). Spe- found in malignant gliomas include broad or regional cifically, p53 signaling was impaired in 87% of the samples alterations spanning segments or whole arms of entire through CDKN2A deletion (49%), MDM2 (14%) and

GENES & DEVELOPMENT 757 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al.

In parallel to TCGA, the Vogelstein laboratory (Parsons et al. 2008) pursued a complementary strategy to charac- terize glioblastoma genomes composed of sequence, copy number, and expression analysis of a greater number of genes in fewer tumors—specifically, 23,219 transcripts from 20,661 -coding genes in 22 malignant glio- mas were analyzed by Sanger-based sequencing. Strik- ingly, five out of 22 tumors, which included one ‘‘high- grade’’ glioma and one secondary glioblastoma, harbored recurrent R132H-encoded substitutions in the isocitrate dehydrogenase 1 (IDH1) . These findings were extended to 18 out of 149 malignant gliomas, which included mostly primary glioblastomas but also a proportion of secondary tumors. Three IDH isoforms exist in humans: IDH1 is a cytosolic protein, whereas IDH2 and IDH3 are located within the mitochondria. IDH1 and IDH2 are known to catalyze the oxidative decarboxylation of isocitrate to a-ketoglutarate (a-KG), leading to the production of NADPH, in the tricarboxylic acid (TCA) cycle, a bio- chemical sequence critical in sugar, lipid, and amino acid metabolism (Raimundo et al. 2011). Although a mis- sense mutation in IDH1, encoding IDH1 R132C, was first identified in one patient with colon cancer in a se- Figure 1. Genomic alterations underlying gliomagenesis. Both quencing analysis of the coding regions in breast and colon primary and secondary glioblastomas arise from precursor cells cancer (Sjoblom et al. 2006), this study provided the first that may be distinct. Primary glioblastomas arise de novo and evidence of recurrent mutations in IDH1. exhibit p53 and Rb pathway dysfunction as well as RTK/Ras/ Subsequent work from several groups has provided PI3K signaling dysregulation, leading to tumors that arise in a comprehensive picture of the IDH status in brain older patients with a worse prognosis, likely owing to the pre- tumors (Balss et al. 2008; Hartmann et al. 2009; Ichimura dominant wild-type IDH1 genotype. In contrast, secondary glioblastomas are preceded by lower-grade II lesions, which et al. 2009; Yan et al. 2009). IDH1/2 is mutated in grade II progress either through grade III lesions or directly to glioblas- and III gliomas as well the secondary glioblastomas that toma. These tumors occur in younger patients and are domi- arise from prior low-grade tumors, with most mutations nated by a mutant IDH1 genotype that confers a better prognosis found in the IDH1 gene. Importantly, these mutations and is associated with a more restricted frontal lobe location. usually occur at conserved residues and are virtually never homozygous. Specifically, whereas only 3%–7% of primary glioblastomas harbor IDH1 mutations, the MDM4 (7%) amplification, and mutation and deletion of majority (50%–80%) of secondary glioblastomas express TP53 (35%). Likewise, Rb signaling was impaired in 78% mutant IDH1. Furthermore, most lower-grade gliomas of the samples through CDKN2 family deletion; amplifi- harbor IDH1 mutations; although grade I pilocytic astro- cation of CDK4 (18%), CDK6 (1%), and CCND2 (2%); and cytomas usually express wild-type IDH1, ;60%–80% mutation or deletion of RB1 (11%). Additional work also of grade II and III astrocytomas, oligodendrogliomas, and showed that CDK6 is an , and particular patterns oligoastrocytomas express mutant IDH1, with the R132H of CDKN2 loss can predict cellular dependency on CDK4 mutation representing the majority of mutations observed. and CDK6 inhibition (Wiedemeyer et al. 2010). Finally, In addition, ;3% of these tumors that express wild-type evidence of RTK/RAS/PI3K activation was found in 88% IDH1 were found to express IDH2 R172 mutations (Balss of tumors, including contributions from unexpected mu- et al. 2008; Hartmann et al. 2009; Ichimura et al. 2009; Yan tations or deletions in NF1 (18%) and PIK3R1,which et al. 2009), although this mutation in IDH2 has only been encodes the p85a regulatory subunit of PIK3CA. Building documented in a single glioblastoma in the literature on this pilot glioblastoma effort, TCGA is profiling pri- (Hartmann et al. 2010). Other CNS tumors found to harbor mary glioblastoma samples with the goal of achieving IDH1 mutations include gangliogliomas, giant cell glio- complete genomic characterization of >500 tumors. As blastomas, and primitive neuroectodermal tumors, although next-generation sequencing technologies (Meyerson et al. small numbers of these tumors have been studied (Balss 2010) are incorporated into the TCGA pipeline, many of et al. 2008). Whereas mutations in other TCA cycle en- these glioblastoma specimens will undergo whole-exome, zymes, such as fumarate hydratase in leiomyomas and renal transcriptome, and/or whole-genome sequencing, providing cell cancer and succinate dehydrogenase in paragangliomas, higher-resolution detail of lower-frequency somatic alter- have been identified (Kaelin 2011; Raimundo et al. 2011), ations, including intrachromosomal and interchromosomal mutations in these genes have not been found in gliomas. translocations. A subset of these data from nearly 500 IDH1/2 mutations have also been identified in 12%– tumors is available through the TCGA portal (http:// 17% of acute myeloid leukemias (AMLs) (Mardis et al. tcga-data.nci.nih.gov/tcga). 2009; Paschka et al. 2010; Ward et al. 2010; Graubert and

758 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma

Mardis 2011), the majority of central and periosteal related HIST1H3B gene, which encodes the H3.1 protein cartilaginous tumors (Amary et al. 2011a), and also 23% isoform, in pediatric diffuse pontine gliomas. Mutations of cholangiocarcinomas (Borger et al. 2012). Interestingly, in these two genes were found in 78% of these tumors, somatic mosaic IDH1/2 mutations were found to be the 22% of nonbrainstem pediatric glioblastomas, and virtu- likely genetic basis of Ollier disease and Maffuci syn- ally no other CNS tumors evaluated. Together, these drome (Amary et al. 2011b; Pansuriya et al. 2011). These findings have clear implications for taxonomic classifi- rare, nonfamilial conditions, both characterized by the cation and, in the case of BRAF alterations, potential early development of multiple cartilaginous tumors, have targeted therapies. also been reported to manifest concomitant glioma or AML, thereby providing an intriguing demonstration of the likely causal role that mutant IDH1/2 plays in these Transcriptional profiling: identification of subtypes three distinct tumor types (Rawlings et al. 1987). Fi- and biological programs in malignant glioma nally, 50% of patients with D-2-hydroxyglutaric aciduria Classification (D-2-HGA), a rare inherited neurometabolic disorder, have been found to carry IDH2 mutations (Kranendijk et al. The genome-wide analysis of mRNA expression to iden- 2010). Indeed, the discovery of IDH1/2 mutations is one tify molecular subclasses (Golub et al. 1999) has led to of the major novel findings to emerge from genome anno- a fundamental shift in our understanding of glioblastoma tation studies and has stimulated renewed attention to subtypes. Indeed, the identification of multiple subtypes altered metabolism in cancer biology. within glioblastoma has underscored the heterogeneity of diseases that all share the same WHO histopathological grade. This approach has revealed that the glioma tran- The genetic basis of oligodendrogliomas scriptome is highly structured and reflects tumor histol- and pediatric gliomas ogy, molecular alterations, and clinical outcome (Nutt In addition to the frequency of IDH1 mutations in grade II et al. 2003; Freije et al. 2004; Phillips et al. 2006; Verhaak glioma, cancer sequencing studies have provided new et al. 2010; Brennan 2011; Huse et al. 2011). insights into the genetic basis of other lower-grade glial Although expression profiling of glioblastoma has been neoplasms. Specifically, >50% of oligodendrogliomas dis- used by many groups (for review, see Brennan 2011; Huse play loss of heterozygosity (LOH) at chromosomes 1p and et al. 2011), two studies have provided the foundation for 19q (Cairncross et al. 1998), although the targets of these classification of glioblastoma subtypes (Phillips et al. 2006; deletions have remained elusive. However, Bettegowda Verhaak et al. 2010). Using unsupervised hierarchical et al. (2011) recently used next-generation sequencing to clustering analysis, Verhaak et al. (2010) classifed 200 analyze the exomes of seven anaplastic oligodendroglio- TCGA glioblastoma samples into four subtypes (Table mas (WHO grade III) and found novel recurrent inactivat- 1), which were subsequently validated using previously ing mutations affecting FUBP1 (far-upstream element published data from 260 independent samples. Each of [FUSE]-binding protein 1; five out of 34 tumors), a regulator the four subtypes was ultimately defined by a mini- of MYC signaling located on chromosome 1p, and the mum list of 210 genes (http://tcga-data.nci.nih.gov/docs/ homolog of Drosopila capicua, CIC (18 out of 34 tumors), publications/gbm_exp). By incorporating the available a downstream transcriptional repressor of RTK/MAPK copy number and sequence data, three of the four sub- signaling located on chromosome 19q. Yip et al. (2012) types were found to harbor distinct molecular alterations. confirmed the high incidence of CIC mutation with con- Specifically, the proneural subtype was enriched for am- current 1p/19q loss and IDH1 mutation in their series. plifications of PDGFRA, CDK6, CDK4,andMET;11outof Recent work has also identified a high incidence of 12 IDH1 mutations found in the TCGA samples; PIK3CA/ specific mutations in two types of pediatric gliomas. First, PIK3R1 mutations; and mutation or LOH of TP53.Ofnote, several studies have revealed BRAF alterations in lower- this subtype contained the highest percentage of young grade pediatric tumors. Copy number analysis of WHO patients, likely due in part to the high number of IDH1 grade I pilocytic astroctyomas identified a tandem dupli- mutant tumors in this category. The classical subtype was cation at chromosome 7q34 resulting in a novel onco- enriched for amplification of EGFR and loss of PTEN and genic BRAF fusion gene, KIAA1549:BRAF,in>60% of CDKN2A, whereas the mesenchymal subtype harbored these tumors (Bar et al. 2008; Jones et al. 2008; Pfister mutations and/or loss of NF1, TP53,andCDKN2A.To et al. 2008). Together with other identified fusion events date, no unique genetic alterations define the neural class such as SRGAP3:RAF1 (Jones et al. 2009), RAF fusion from the other classes. events occur in >80% of pilocytic astrocytomas (von In contrast to Verhaak et al. (2010), Phillips et al. (2006) Deimling et al. 2011). Furthermore, BRAFV600E muta- identified three distinct glioblastoma subtypes based on tions have been found most commonly in WHO grade II their differences of expression of a panel of genes most pleomorphic xanthoastrocytomas (66%) (Dias-Santagata strongly correlated with survival. This approach delin- et al. 2011; Schindler et al. 2011) as well as WHO grade I eated subtypes that were termed proneural, mesenchy- gangliogliomas (18%) (MacConaill et al. 2009; Schindler mal, and proliferative. Although the number of subtypes et al. 2011). In addition, Wu et al. (2012) used whole- identified by the Verhaak et al. (2010) and Phillips et al. genome sequencing to identify recurrent mutations in (2006) studies differs, the proneural and mesenchymal H3FA, which encodes the H3.3 protein, and the closely classifications identified using distinct methodologies

GENES & DEVELOPMENT 759 6 GENES 760 al. et Dunn Downloaded from & DEVELOPMENT genesdev.cshlp.org

Table 1. Molecular heterogeneity and classification in gliomas/astrocytomas Secondary glioblastoma Glioma Oligodendroglioma DIPG (III–IV) PA, PXA(I) Primary glioblastoma (IV) (IV) (II–III) (II–III) (pediatric) (pediatric)

Transcriptional Mesenchymal Classical Neural Proneural onSeptember30,2021-Publishedby subtype IDH1/2 status Wild type Wild type Wild type Mutant Mutant Mutant Mutant Methylation feature G-CIMP+ G-CIMP+ G-CIMP+ G-CIMP+ Signature mutations NF1 TP53 TP53 TP53 CIC H3F3A, BRAFV600E (PXA), HIST1H3B BRAF fusions (PA) Signature CNAs NF1 loss PTEN loss PDGFRA amp PDGFRA 1p/19q loss EGFR amp MET amp amp Proteomic features NF1 signaling, EGFR signaling, PDGFB signaling YKL-40, VEGF, Notch signaling CD44, IRS1 Transcriptional STAT3, CEBP/b, regulators TAZ Cold SpringHarborLaboratoryPress Primary glioblastomas are characterized by particular transcriptomic subtypes that are further described by specific genomic and proteomic features. Secondary glioblastomas as well as lower-grade gliomas also exhibit signature genomic alterations. (PA) Pilocytic astrocytoma; (PXA) pilocytic xanthoastrocytoma; (DIPG) diffuse intrinsinc pontine glioma (DIPG).

Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma and sample sets are the most robust and concordant (Huse (Basso et al. 2005) to identify transcription factors that et al. 2011); genes defining the Verhaak and Phillips sub- regulate the mesenchymal expression state (Carro et al. classes are listed together in Supplemental Table S1B of 2010; Bhat et al. 2011). Using ARACNe and a novel Bhat et al. (2011,). For instance, both groups identified master regulatory algorithm, Carro et al. (2010) identi- proneural class expression of DLL3 and OLIG2 and mes- fied six transcription factors—STAT3, C/EBPb,RUNX1, enchymal class expression of CD40 and CHI3L1/YKL-40, FOSL, bHLH-B2, and ZNF-238—that are likely to control the latter of which appears to be a potential serum protein the majority of mesenchymal subtype tumors. Overex- marker of prognosis in glioblastoma patients (Iwamoto pression of STAT3 and C/EBPb both induced a mesenchy- et al. 2011). Moreover, a subset of the genes represented mal phenotype and were individually required for ortho- in these subtypes is represented in a nine-gene panel shown topic xenograft growth. Bhat et al. (2011) also applied the to predict outcome in glioblastoma, as increased expres- ARACNe method to TCGA samples to identify transcrip- sion of mesenchymal genes such as CHI3L1/YKL-40 tion factors that regulated the mesenchymal subtype and and LGALS3 combined with decreased expression of a identified several distinct components, including TAZ, proneural gene, OLIG2, were associated with typical YAP, MAFB, and HCLS1. Follow-up work demonstrated short-term survival compared with longer-term survivors a pivotal involvement of TAZ and dysregulated Hippo (Colman et al. 2010). Considering that other groups have pathway signaling in driving mesenchymal biology; TAZ used alternative methodologies to derive nonoverlapping tended to be hypermethylated in proneural tumors, could prognostically significant gene sets (Bredel et al. 2009), drive a mesenchymal phenotype when overexpressed, and further work in large, relatively uniform sample sets will cooperated with PDGF-B to induce malignant mesenchy- be necessary to resolve subtype classification systems mal-type gliomas in the RCAS/N-tva mouse model. To- and validate multiple prognostic gene sets. gether, these network analyses of data To determine whether these genomically identified demonstrate the utility of these data sets not only to subtypes predict biological or clinical differences in glio- classify distinct tumor states, but also to provide clues to blastoma, Brennan et al. (2009) used a targeted proteomics the underlying biology. approach to determine whether glioblastomas also segre- gated into distinct classes by activation of signal trans- duction pathways. Unsupervised clustering of 57 Signature molecular lesions in glioma: key drivers, or protein modifications assessed in 20 glioblastoma tumor suppressors, and the IDH proteins samples identified distinct tumor subgroups defined by Combined work from genomic and proteomic analyses EGFR-related signaling, PDGF-related signaling, or pro- has focused attention on critical pathways—driven by the teins associated with decreased NF1 expression. More- dysregulation of driver and checkpoint proteins—that over, analysis of TCGA glioblastoma expression profiles contribute to gliomagenesis (Phillips et al. 2006; The showed that SCNAs or mutations of EGFR, PDGFRA, Cancer Genome Atlas Research Network 2008; Parsons and NF1 were largely mutually exclusive, suggesting that et al. 2008; Brennan et al. 2009; Verhaak et al. 2010). the identified signaling nodes are nonoverlapping. Inter- In particular, it is clear that RTKs (including EGFR, estingly, because PDGF core tumors often expressed high PDGFRA, and MET), the PI3K pathway, signaling path- PDGF protein but low PDGF mRNA and no PDGFRA ways activated by PTEN and NF1 loss, and the mutant SCNAs, it is possible that PDGF/PDGFRA-mediated bi- IDH proteins play central roles in the pathobiology of ology is underestimated in purely mRNA-based expres- glioblastoma. sion schemes. Moreover, the finding that EGFR core tumors harbored evidence of active Notch signaling not EGFR/EGFRvIII reflected in copy number, expression, or sequence data was additional evidence that not all biologically impor- EGFR amplification is observed in ;50% of primary tant information can be gleaned from genomic data. This glioblastomas and is associated with poor prognosis study supports the integration of protein-based biomarker (Hurtt et al. 1992; Jaros et al. 1992; Schlegel et al. 1994). assessment, perhaps by immunohistochemistry for a tar- Furthermore, ;50% of EGFR-amplified cells harbor the geted biomarker panel, into the broader subtype classifi- EGFRvIII mutant, which is an intragenic gene rearrange- cation architecture currently defined by genomic profiling. ment generated by an in-frame deletion of exons 2–7 that encode part of the extracellular region. The expression of EGFRvIII has been determined to confer a worse progno- Network analysis of expression profiles sis than wild-type EGFR expression alone (Shinojima Additional work has extended the utility of mRNA pro- et al. 2003; Heimberger et al. 2005). Experimentally, filing by using computational network analysis to un- ectopic overexpression of EGFRvIII in glioma cell lines cover the causal regulatory modules underlying particu- induces constitutive autophosphorylation, activation of lar transcriptomically defined subtypes. Broadly, these the Shc–Grb2–Ras and class I PI3K pathways (Huang et al. approaches attempt to infer the upstream biological pro- 1997; Narita et al. 2002), enhanced tumorigenicity (Huang grams, or networks, that coordinately produce a given et al. 1997), increased cell proliferation (Narita et al. 2002), transcriptional state. Recently, two groups used a reverse and resistance to apoptosis induced by DNA-damaging engineering-based algorithm called ARACNe (algorithm agents through modulation of Bcl-XL expression (Nagane for the reconstruction of accurate cellular networks) et al. 1998). Notably, the downstream effects of EGFRvIII

GENES & DEVELOPMENT 761 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al. overexpresson are not recapitulated by overexpression of well as in vitro when grown in stem cell culture condi- wild-type EGFR. For example, wild-type EGFR cannot tions (Stockhausen et al. 2011), suggesting that durable substitute for EGFRvIII in driving infiltrative glioma EGFRvIII expression may be linked to differentiation and/ formation in genetically engineered mice (Hesselager and or development. Moreover, it is now clear that the type of Holland 2003; Zhu et al. 2009) or in Ink4a/ArfÀ/À murine genetic alterations involving EGFR in glioblastoma are neural stem cells or astrocytes (Holland et al. 1998; distinct from those observed in other EGFR-altered cancers, Bachoo et al. 2002), except when EGF is infused such as non-small-cell lung cancer (NSCLC). In glioma, at a high concentration into the injection site of wild- focal EGFR amplification occurs at an extremely high level type EGFR-transduced cells (Bachoo et al. 2002). Both (>20 copies). In addition, the vast majority of other muta- EGFRvIII and wild-type EGFR/ErbB family proteins have tions, including the vIII mutant as well as missense been identified in the nucleus and are thought to drive mutations (Lee et al. 2006b; The Cancer Genome Atlas proliferation and DNA damage repair through both tran- Research Network 2008), are found within the extracel- scriptional and signaling functions (Wang and Hung lular domain, while most mutations in other nonglioma 2009). Moreover, the observation that EGFR also trans- cancers are found in the intracellular domain (Janne et al. locates to the mitochondria (Boerner et al. 2004) provides 2005). Although EGFRvIII expression is sufficient to further evidence that the contributions of EGFR malig- rescue the knockdown of an endogenous kinase domain nancy may not be limited to its conventional cell mem- mutant EGFR in NSCLC cells (Rothenberg et al. 2008), it brane location and merit further study. is not clear whether EGFR mutant glioma cells drive Despite the well-recognized proproliferative functions similar downstream signaling and/or confer the same of EGFRvIII, its expression in human glioblastoma is ‘‘addiction’’ to EGFR activation as is the case in NSCLC heterogeneous and is most often observed only in a sub- (Sharma and Settleman 2007). population of cells (Fig. 2A; Nishikawa et al. 2004). Recent observations support a model of functional hetero- PDGFR geneity in which a minority of EGFRvIII-expressing cells not only drive their own intrinsic growth, but also potentiate Nearly 30% of human gliomas show expression patterns the proliferation of adjacent wild-type EGFR-expressing cells that are correlated with PDGFR signaling (Brennan et al. in a paracrine fashion through the cytokine coreceptor 2009) and genes involved in oligodendrocyte develop- gp130 (Inda et al. 2010). Even though these results illus- ment (OLIG2, NKX2-2, and PDGF), both hallmarks of trate that cytokines produced from EGFRvIII expression the proneural glioblastoma subtype. PDGFRA amplifica- can be drivers of heterogeneity, there are likely additional tion is found in 15% of all tumors and is enriched in the cytokine-inducing mechanisms at work. For example, it proneural subtype (Phillips et al. 2006; Verhaak et al. has been recently shown that NFKB1A, which encodes 2010). Of those tumors harboring gene amplification, IkBa, a critical negative regulator of canonical NF-kBacti- recent work showed that 40% harbor an intragenic de- vation, was found to undergo monoallelic loss in glioblas- letion, termed PDGFRAD8,9 (Clarke and Dirks 2003), in tomas that lack EGFR amplification (Bredel et al. 2011), which an in-frame deletion of 243 base pairs (bp) of exons suggesting that NF-kB plays physiologically relevant roles 8 and 9 leads to a truncated extracellular domain (Ozawa downstream from EGFR/EGFRvIII that include IL-8 pro- et al. 2010). In addition, in-frame gene fusion of the duction (Bonavia et al. 2011). These results suggest that extracellular domain of KDR/VEGFR-2 and the kinase intraclonal drives the persistence of intra- and intracellular domains of PDGFRA has also been tumoral heterogeneity, which has implications for both identified, and both the PDGFRAD8,9 and KDR-PDGFRA our basic understanding of gliomagenesis and also drug mutant proteins were constitutively active and trans- sensitivity profiles of these tumors (Yao et al. 2010). forming and could be inhibited with inhibitors of PDGFRA. In addition, although most human glioma cell lines fail Point mutations in PDGFRA are associated with amplifi- to faithfully recapitulate the EGFR amplification and cation but, unlike EGFR, are generally rare events (The EGFRvIII expression observed in primary tumor speci- Cancer Genome Atlas Research Network 2008). Of the mens, recent studies have reported successful passage of additional ways to activate PDGFR signaling, PDGF li- EGFRvIII-expressing glioblastoma xenografts in vivo as gands (A–D) are up-regulated in ;30% of glioma surgical

Figure 2. Molecular heterogeneity in glioblastoma. (A) Immunohistochemistry for the mutant EGFR receptor EGFRvIII demonstrates a heterogeneous staining pattern within the tumor. Images from Nishikawa et al. (2004) used with permission. (B) Multicolor FISH reveals distinct subpopulations of either EGFR (red) or PDGFRA (green) amplifica- tion within a glioblastoma specimen. Images obtained from Cameron Brennan.

762 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma samples and cell lines, while PDGFRB expression appears HGF (Pillay et al. 2009). Similar results were achieved to be restricted to proliferating endothelial cells in glio- with combined c-Met and EGFR small molecule inhibitors blastoma (Fleming et al. 1992; Hermanson et al. 1992; Di (Huang et al. 2007; Stommel et al. 2007) and neutralizing Rocco et al. 1998; Smith et al. 2000; Lokker et al. 2002). anti-HGF monoclonal antibody combined with erlotinib The intratumoral coexpression of PDGF and PDGFR (Lal et al. 2009). The relationship of c-Met with EGFR is suggests that both autocrine and paracrine loops play also not surprising in light of prior studies showing that roles in glioblastoma. This possibility is supported by the overexpression of Met can lead to gefitinib resistance in demonstration that stimulation of PDGFRB-expressing mutant EGFR lung cancers, often by activating ERBB3 endothelial cells by tumor-derived PDGF can drive VEGF- (Engelman et al. 2007). Concurrent activation of c-Met mediated angiogenesis within the local microenvironment with PDGFR has also been detected in glioblastoma and (Guo et al. 2003). has been suspected to be another mechanism for resistance Similar to the case of EGFR/EGFRvIII described above, to EGFR kinase inhibitors. In this case, inhibition of three cell and receptor heterogeneities appear to be other dis- RTKS—EGFR inhibition with erlotinib, c-Met inhibition tinct mechanisms by which PDGF/PDGFR promotes with SU11274, and PDGFR inhibition with imatinib— aggressive glioma growth. For example, transduction of significantly inhibited the in vitro growth of glioblastoma cells of the subventricular zone (SVZ) of the lateral cell lines, compared with single drugs alone (Stommel ventricle of neonatal rat pups with a retrovirus express- et al. 2007), possibly by attenuating downstream PI3K ing PDGF yielded large, diffusely infiltrating tumors signaling. In contrast, in scenarios in which MET is resembling glioblastoma (Assanah et al. 2006, 2009). amplified in isolation, one report showed that treatment The tumors that formed contained a massive prolifera- with crizotinib, which inhibits ALK and also c-Met tion of both infected and uninfected PDGFRa+-express- (Christensen et al. 2007), can induce radiographic and ing progenitors, suggesting that PDGF was driving tumor clinical improvement (Chi et al. 2012). formation through both autocrine and paracrine signaling, leading to the recruitment of non-PDGF-expressing resi- Src family (SFKs) dent progenitors and the evolution of cellularly heteroge- SRC and SFKs are frequently activated in glioblastoma neous malignant gliomas. Further work has demonstrated patient samples and cell lines (Stettner et al. 2005; Du that these recruited cells become transformed, overtake et al. 2009) and are widely expressed in glioblastoma (Lu PDGF-induced gliomas, and can be serially transplanted et al. 2009). SFKs mediate signaling from (Fomchenko et al. 2011), and that topotecan delivery to receptors that are commonly overexpressed in glioblas- these tumors results in ablation of both tumor-initiating toma, providing a potential explanation for SFK activa- cells and recruited glial progenitors (Lopez et al. 2011). tion. Bead-based profiling of activation in Although these results still remain to be validated within 130 human cancer cells showed that the most frequently the context of human tumors, they raise the possibility activated tyrosine kinases were EGFR, fibroblast growth that cells distinct from the initial transformed cells of factor receptor 3 (FGFR3), protein (PTK2, origin within the tumor environment can be corrupted to also known as kinase, or FAK), and SFKs become bona fide tumor cells. This model of glioma including SRC, LYN, and (Du et al. 2009). Moreover, evolution is distinct from the generally established view screening of 31 primary glioblastomas samples showed of linear gliomagenesis (Fomchenko et al. 2011). SRC activation in 61% of samples (Du et al. 2009) and that the SRC inhibitor dasatinib inhibited cell viability and c-Met migration in vitro and tumor growth in vivo, nominating SRC/SFK as potential therapeutic targets in a subset of The c-Met RTK is amplified in ;5% of glioblastomas, has glioblastomas. been found to be overexpressed in 18 out of 62 (29%) glioblastoma samples with shorter median survival (Kong RTK cooperativity and heterogeneity et al. 2009), and is rarely mutated (The Cancer Genome Atlas Research Network 2008). Moreover, c-MET was One potential explanation for the failure of EGFR and found to be coactivated in glioblastoma cells with in- PDGFRA inhibitors to elicit significant clinical outcomes creased levels of EGFR/EGFRvIII (Huang et al. 2007; (De Witt Hamer 2010) is that additional RTKs may co- Stommel et al. 2007; Pillay et al. 2009) and represents operate to provide an integrated signaling threshold that is a critical dependency in MET-amplified cells (Beroukhim not sufficiently attenuated through the inactivation of any et al. 2007). In this setting, activated EGFR can associate single RTK (Huang et al. 2007; Stommel et al. 2007). with c-Met, leading to the activation of this receptor in Indeed, Stommel et al. (2007) demonstrated that three or the absence of its ligand, HGF (Jo et al. 2000). Conversely, more RTKs were activated in a majority of glioblastoma HGF transcriptionally activates the expression of the cell lines and patient specimens. This discovery of con- EGFR ligands TGF-a- and heparin-binding EGF and can comitant receptor expression and coactivation suggests therefore activate EGFR (Reznik et al. 2008). Blocking that tumor RTK profiling may be an important step in the EGFRvIII activity with an EGFR-specific monoclonal development of a personalized glioblastoma therapeutic antibody, panitumumab, can result in a switch to HGF- regimen and that cross-talk between the receptors could mediated c-Met activation, which could be prevented by be targeted with specific inhibitors to both, resulting in cotreatment with AMG102, a neutralizing antibody to enhanced cytotoxicity.

GENES & DEVELOPMENT 763 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al.

Recent studies provide additional evidence that glio- by PDK1, and Ser 473 phosphorylation blastomas are composed of heterogeneous subpopula- by the mammalian target of rapamycin (mTOR) complex tions within tumors. Indeed, varied expression patterns 2 (mTORC2) (Sarbassov et al. 2005). Elevated AKT phos- of several proteins have been described: Wild-type EGFR phorylation has been observed in up to 85% of glioblas- and EGFRvIII, as described above (Nishikawa et al. 2004; toma cell lines and patient samples (Wang et al. 2004). Inda et al. 2010); PDGFRA (Hermanson et al. 1992); RTK-independent activation of this pathway in glioblas- c-Met (Nabeshima et al. 1997); angiogenic factors (Koga toma can occur via mutation or amplification of PIK3CA et al. 2001); and adhesion molecules (Bello et al. 2001b) (p110a) (Gallia et al. 2006; Kita et al. 2007; The Cancer all exhibit heterogeneous distributions. Two recent studies Genome Atlas Research Network 2008), and PIK3CD provide further compelling examples of clonal heteroge- (p110d) is also overexpressed in some gliomas (Mizoguchi neity in glioblastoma. Both Snuderl et al. (2011) and Szerlip et al. 2004). Moreover, the TCGA study revealed recurrent et al. (2012) observed that 5%–7% of all glioblastomas—and mutations in the gene encoding the p85a regulatory subunit nearly 13% of glioblastomas with EGFR, PDGFRA,orMET PIK3R1, which likely drive PIK3CA activation through amplification—harbored multiple RTK amplifications. decreased SH2 domain-mediated inhibition (Sun et al. 2010). Moreover, while a minority of cells harbored amplifica- PI3K drives many glioma-relevant processes, including tion of multiple RTKs (Szerlip et al. 2012), the pre- survival, proliferation, migration, and invasion (Engelman dominant pattern was mosaic amplification such that et al. 2006). In addition to its well-known functions (Lino tumors were comprised of individual cell populations and Merlo 2011), recent work has revealed several novel harboring isolated amplification of a single RTK (Fig. mechanisms underlying PI3K function in glioma. Specifi- 2B). This point was further demonstrated in a remarkable cally, CD95 promotes invasion through interaction with FISH analysis of tumor sections from a whole-brain PI3K and the SFK Yes, triggering activation of GSK3b and autopsy of an untreated patient with bilateral multifocal induction of matrix metalloproteinase (MMP) expression glioblastoma in which there was a striking anatomic (Kleber et al. 2008). The RTK EphA2, which is highly distribution of EGFR-andPDGFRA-amplified cells expressed in glioblastoma (Wykosky et al. 2005) and binds (Snuderl et al. 2011). It is likely that this pattern of clonal PI3K upon ligand stimulation (Pandey et al. 1994), induces RTK heterogeneity is a late event in gliomagenesis glioblastoma cell migration in an AKT-dependent manner (Snuderl et al. 2011; Szerlip et al. 2012) and is influenced (Miao et al. 2009). In addition, the insulin-like growth by several factors, including binomial segregation of un- factor 2 (IGF2) promotes aggressive growth in glioblastoma stable amplicons or local microenvironment selection lacking EGFR amplification or overexpression through (Szerlip et al. 2012). It is likely that somatic mutations in IGF receptor 1 (IGFR1) and PIK3R3 (Soroceanu et al. cancer genes will demonstrate a pattern of heterogeneity 2007). The PI3K pathway also maintains glioblastoma similar to that seen with SCNAs, but this possibility tumor-initiating cells, as direct silencing of mTOR or awaits additional study at this point. Together, these inactivation with rapamycin reduces neurosphere for- studies provide further molecular evidence that glioblas- mation and expression of neural stem cell progenitor tomas are often heterogeneous, which has major biologic markers (Sunayama et al. 2010) as well as growth of and therapeutic implications. patient-derived tumor-initiating cells (Gallia et al. 2009). In addition, AKT activation due to PTEN loss likely contributes to RTK inhibitor insensitivity in glioblastoma PI3K pathway (Mellinghoff et al. 2005, 2007). Activating kinases The PI3K signaling pathway is dys- Effectively inhibiting the PI3K signaling pathway is regulated in many cancers (Yuan and Cantley 2008), challenging because the cascade and its feedback regula- including glioblastomas. Some of the major genomic tion remain incompletely understood. Specifically, RTK- alterations discussed above—RTK amplification/muta- activated PI3K signaling does not always require AKT, as tion, PIK3CA and PIK3R1 mutation, and PTEN loss—all EGFR can also signal to mTORC1 through activate this pathway (The Cancer Genome Atlas Re- C (PKC) independently of AKT (Fan et al. 2009), perhaps search Network 2008). Of the three PI3K classes, the class through PDK1 (Dutil et al. 1998; Le Good et al. 1998). IA kinases are likely to play a direct role in cell trans- Indeed, a novel PI3K/PDK1/PKCi pathway can regulate formation and are composed of both a catalytic subunit phosphorylation and inactivation of the proapoptotic pro- isoform (p110a, p110b, p110d, and p110g) and a regulatory tein Bad, thereby increasing glioma cell survival (Desai et al. subunit isoform (p85a,p55a, p50a, p85b, and p55g). 2011). Additional work has revealed the existence of an Broadly, activation of the pathway can be initiated by GTP- AKT-independent, mutant PIK3CA signaling pathway lead- bound Ras (Rodriguez-Viciana et al. 1994, 1996) and through ing to PDK1-mediated activation of a critical downstream RTK signaling, which recruits PI3K to the cell membrane, effector, SGK3, in PTEN intact tumor cells (Vasudevan whereupon the lipid phosphatidylinositol (PtdIns)-4,5- et al. 2009). PI3K pathway inhibition is thought be bisphosphate (PIP2) is phosphorylated to PtdIns-3,4,5- cytostatic, rather than cytotoxic, potentially due to G1 bisphosphate (PIP3). This reaction is antagonized by the cell cycle arrest (Paternot and Roger 2009; Fan et al. major glioma tumor suppressor and PtdIns(3,4,5)P3 phos- 2010). Combination mTOR and MEK inhibition has phatase PTEN. PIP3 subsequently recruits the serine/ therefore been used that suppresses CDK4 phosphoryla- threonine kinase AKT to the plasma membrane, where tion in a synergistic manner (Paternot and Roger 2009). it is fully activated by its main dual inputs—Thr 308 Due to complex feedback pathways, other combination

764 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma therapies being explored to target PI3K signaling in grade III astrocytomas and progression into glioblastoma glioma include inhibition of PI3K and mTOR with in- (Kwon et al. 2008). Together, these models demonstrate hibitors such as PI-103 (Fan et al. 2006), as well as the causal role of PTEN loss in glioblastoma and provide simultaneous inhibition of mTORC1 and mTORC2 (Q the means with which to study patient-relevant disease Liu et al. 2011). Monotherapy with the mTORC1 in- in tractable murine models. hibitor rapacmycin disrupts an IRS-1-mediated negative Clinically, PTEN loss has been shown to confer resis- feedback loop and can actually increase AKT activity tance to EGFR inhibitors in patients harboring EGFRvIII- (Fan et al. 2006; Cloughesy et al. 2008) through mTORC2- expressing glioblastoma in part due to its activation of mediated phosphorylation (Gulati et al. 2009). Interest- downstream AKT (Mellinghoff et al. 2005) as well as loss ingly, dual PI3K/mTOR inhibition with PI-103 induces of its RTK degradation function (Vivanco et al. 2010). autophagy, which, when also inhibited pharmacologi- PTEN has also been implicated in the ubiquitin-mediated cally, leads to apoptosis (Fan et al. 2010). Thus, a com- control of protein stability with respect to apoptosis, as it binatorial therapeutic strategy for targeting the PI3K influences the half-life of the anti-apoptotic protein FLIPS pathway in glioblastoma will likely be necessary as (Panner et al. 2009). With respect to development, PTEN more of its complex biology and regulation are revealed loss may also promote a ‘‘side population’’ phenotype of (Akhavan et al. 2010). glioma stem-like cells by driving expression of the drug transporter protein ABCG2 (Bleau et al. 2009). Further- PTEN PTEN directly antagonizes PI3K signaling and is more, recent studies have described new roles for PTEN one of the most frequently altered genes in cancer. It in metabolism and cellular homeostasis; in PTEN-null undergoes genomic loss, mutation, or epigenetic inacti- cells, the ectonucleoside triphosphate diphospho- vation in 40%–50% of gliomas, resulting in high levels of 5 (ENTPD5) promotes protein N-glycosylation PI3K activity and downstream signaling (Koul 2008). As and folding in the setting of increased AKT-mediated the PI3K pathway is a driving force in gliomas, even small anabolism, increases levels, con- changes in the expression or function of its critical negative tributes to increased aerobic glycolysis (i.e., the Warburg regulator, PTEN, have profound effects on tumor cell effect), and is required for PTEN-null tumor cell growth behavior. The stability of PTEN is regulated post-trans- (Fang et al. 2010), although this result awaits validation in lationally by GSK3-mediated phosphorylation at Thr glioma models. The nuclear localization of PTEN may also 366 (Maccario et al. 2007) and by proteasomal degradation play an important role in tumor suppressor activity (Perren through polyubiquitination by the HECT domain ubiq- et al. 2000; Whiteman et al. 2002; Zhou et al. 2002). uitin NEDD4-1 (Wang et al. 2007). NEDD4-1 up- Specifically, nuclear PTEN interacts with APC/C to pro- regulation has been recently associated with overexpres- mote formation and enhance the tumor-suppressive abil- sion of the FoxM1B transcription factor in gliomas (Dai ity of the APC–CDH1 complex in a phosphatase-indepen- et al. 2007). Moreover, a multiprotein tumor suppressor dent manner (Song et al. 2011). This demonstrates that loss network exists in glioblastoma comprised of PTEN, the and mutation of PTEN are not synonymous, as cells in adaptor protein Na+/H+ exchanger regulatory factor these respective states are differentially sensitive to phar- (NHERF1), and the pleckstrin homology domain leucine- macological inhibition of APC–CDH1 targets. Further rich repeat 1, which form a heterotri- work is necessary to fully understand which of the many meric complex that undergoes disruption in high-grade functions attributed to PTEN play key roles in human tumors (Molina et al. 2012). Thus, in gliomas where PTEN glioblastomas. is not deleted, mutated, or epigenetically silenced, mech- anisms such as aberrant up-regulation of NEDD4-1 or loss NF1 Neurofibromin, the product of the NF1 gene, is of NHERF1 could contribute to suppressing PTEN func- a potent tumor suppressor that negatively regulates Ras tion. These mechanisms raise the possibility that the level and mTOR signaling in astrocytes. Inactivation of NF1 of PTEN dysfunction is underestimated solely on the basis has been observed in gliomas and can arise as a result of of genomic data. excessive proteasomal degradation mediated by hyper- The physiologic relevance of PTEN loss was underscored activation of PKC (The Cancer Genome Atlas Research using several genetically engineered mouse models. CNS- Network 2008; McGillicuddy et al. 2009) or by genetic specific GFAP-Cre-p53lox/loxPTENlox/+ mice developed ma- loss or mutation, which were surprising findings from lignant gliomas with short latency that was strikingly large-scale sequencing analyses (Parsons et al. 2008). NF1 similar to human disease both histopathologically and mutations are most commonly found in the mesenchy- molecularly, given the degree of concomitant RTK acti- mal subtype of glioblastoma (Verhaak et al. 2010). Exper- vation (Zheng et al. 2008). Moreover, dual loss of p53 and iments using NF1-deficient primary murine astrocytes PTEN in this model promoted increased c-Myc activity, have revealed that NF1 loss results in increased cell which led to impaired differentiation and stable tumor- proliferation and migration that is dependent on Ras- igenic capacity of glioma tumor-initiating cells. Mice mediated hyperactivation of mTOR. In this setting, conditionally lacking p53, PTEN, and Rb also formed mTOR induces rapamycin-sensitive activation of Rac1 glioblastoma with SCNAs recapitulating those seen in GTPase that is independent of elongation factor 4E- human tumors (Chow et al. 2011). Furthermore, succes- binding protein 1(4EBP-1)/S6 kinase (S6K) (Sandsmark sive loss of each PTEN allele in the NF1À/À p53À/À model et al. 2007). In addition, NF1 deficiency causes hyper- of progressive astrocytoma accelerated formation of proliferation and impacts glioma formation in a manner

GENES & DEVELOPMENT 765 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al. that is independent of the tuberous sclerosis complex Work on the downstream biological effects of IDH1/2 (TSC)/Ras homolog enriched in brain (Rheb) control of mutation expression has focused largely on the inhibition mTOR (Banerjee et al. 2011). Other potential down- of a-KG-dependent dioxygenases by 2-HG. This diverse stream targets of neurofibromin include Stat3, which group of controls a broad range of physiological was identified in a chemical library screen using NF1- processes, including hypoxic sensing, histone demeth- deficient malignant peripheral nerve sheath tumor cells ylation, demethylation of hypermethylated DNA, fatty (Banerjee et al. 2010). In this case, Stat3 is regulated in an acid metabolism, and collagen modification, among others mTORC1 and Rac1-dependent manner and increases (Loenarz and Schofield 2008). Several studies have provided cyclinD1 expression. Additionally, it will be important evidence to demonstrate that several of these functions are to determine whether NF1-deficient gliomas are as influenced by IDH1/2 mutation expression. IDH1 mu- susceptible to proteoxicity through HSP0/mTOR inhi- tant gliomas exhibit a global DNA hypermethylation bition as in other NF1-deficient cancers (De Raedt et al. state, termed the glioma CpG island methylator pheno- 2011). type (G-CIMP) (Noushmehr et al. 2010). This state was Genetically engineered mouse models have demon- also observed in IDH1/2 mutant AML (Figueroa et al. strated that targeted homozygous loss of NF1 in astro- 2010). Mutant IDH1 expression was sufficient to pro- cytes, while sufficient to increase cell growth in vitro and duce a G-CIMP phenotype in engineered normal human in vivo, is not sufficient to induce glioma formation astrocytes that was highly concordant with that seen in (Bajenaru et al. 2002). Interestingly, NF1À/À astrocytes IDH mutant human tumor samples. Furthermore, this develop optic gliomas in the context of an NF1+/À brain methylation phenotype correlated with a gene expression environment (Bajenaru et al. 2003; Zhu et al. 2005), in signature comprised of a limited set of down-regulated part through paracrine factors including hyaluronidase genes that discriminated between IDH1 mutant and wild- (Daginakatte and Gutmann 2007). In addition, it has been type proneural tumors. This hypermethylation state may found that low levels of cAMP expression in the stroma be caused in part by the 2-HG-mediated inhibition of the are sufficient to induce optic glioma formation in genet- a-KG-dependent TET2 enzyme (Xu et al. 2011; Turcan ically engineered mouse models of NF1 such that local et al. 2012); the resultant decrease in 5-hydroxymethylcy- depletion of cAMP resulted in glioma formation in re- tosine was also observed in glioblastoma specimens (Xu gions of the brain otherwise not observed to develop et al. 2011). Moreover, mutant IDH1/2 cells displayed tumors (Warrington et al. 2010). These findings empha- impaired hematopoietic differentiation, suggesting that size the importance of heterogeneity and the cell type- a hypermethylated epigenetic landscape contributed to specific effects of various genetic alterations in tumori- a persistent dedifferentiated state (Figueroa et al. 2010). genesis. Other genetically engineered mouse models have The inhibition of histone demethylases in IDH1 mutant demonstrated that NF1 loss in glial cells, in combination cells may also impair differentiation (Chowdhury et al. with a germline p53 mutation, results in fully penetrant 2011; Xu et al. 2011; Lu et al. 2012). Repressive histone malignant astrocytomas (Zhu et al. 2005), which progress methylation was shown to be associated with impaired to glioblastoma upon deletion of PTEN (Kwon et al. 2008). mutant IDH1-expressing astrocyte differentiation, and the More recent work has revealed that the same combina- accumulation of these histone marks preceded significant tion of genetic alterations in these tumor suppressor DNA hypermethylation in engineered IDH1 mutant cells genes in neural stem/progenitor cells is necessary and (Lu et al. 2012). Furthermore, Lai et al. (2011) showed that sufficient to induce astrocytoma formation (Alcantara global expression profiles of IDH1 mutant glioblastomas Llaguno et al. 2009). more closely resembled lineage-committed neural precur- sors, whereas wild-type counterparts appear to resemble neural stem cells. Mutant IDH1/2 proteins The production of 2-HG also appears to influence Mechanistic basis of mutant IDH1 biology Mutant HIF biology in several ways. First, 2-HG may stabilize IDH1/2 have been shown to catalyze a neomorphic func- HIF-1 under some conditions (Zhao et al. 2009). How- tion (Dang et al. 2010). Specifically, while wild-type IDH1 ever, the HIF-1 response to hypoxia in IDH mutant cells catalyzes the NADP+-dependent oxidation of isocitrate, is attenuated (Koivunen et al. 2012). Specifically, the mutant IDH1 catalyzes the NADPH-dependent reduction (R)-enantiomer of 2-HG stimulates the EGLN prolyl of a-KG to the (R)-enantiomer of 2-hydroxyglutarate (2- 4-hydroxylases, which mark HIF for degradation. Either HG), which is the same stereoisomer of 2-HG that is seen expression of mutant IDH1, suppression of HIF-1a,or in D-2-HGA. Although wild-type IDH1 can also catalyze overexpression of EGLN1 was sufficient to stimulate this particular reaction (Pietrak et al. 2011), mutant colony formation by immortalized human astrocytes, enzymes perform this reaction with much higher effi- and HIF-regulated genes were down-regulated in IDH1 ciency because the R132 substitutions modify the active mutant proneural tumors. Hypoxic cells drive lipogenesis site to increase a-KG and NADPH binding (Dang et al. via reduction of glutamine to a-KG by wild-type IDH1 2010; Pietrak et al. 2011). Dang et al. (2010) showed that (Metallo et al. 2012). This observation suggests a mecha- mutant cells contained extremely high levels of 2-HG, nism by which cells can survive under hypoxic or pseu- which was also found in primary IDH1 mutant gliomas dohypoxic conditions and points to a physiological selec- and in the serum of IDH mutant AML patients (Gross et al. tion pressure to maintain a copy of the wild-type IDH1 2010; Ward et al. 2010). gene in cancer cells. Finally, mutant IDH1/2 expression or

766 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma

2-HG administration can potently inhibit multiple histone et al. 2009) or its 2-HG by-product (Sahm et al. 2012) in demethylases, likely altering transcriptional programs. clinical settings. While 2-HG is easily detected in the Additional work has shown that cells expressing the serum of AML patients, the correlation between serum IDH1 R132H mutation display metabolomics alterations 2-HG and tumor mutation status may be less specific in in amino-free and branched chain amino acid levels and the setting of glioma (Capper et al. 2011). Finally, it may also also choline phospholipid synthesis (Reitman et al. 2011). be possible to monitor the presence of 2-HG noninvasively Thus, together, these data have pointed to several bi- using magnetic resonance spectroscopy (MRS) imaging of ological processes affected by mutant IDH1/2 expression the brain (Pope et al. 2012). Together, these studies provide that, collectively, may promote tumor growth by integrat- overwhelming evidence for the clinical relevance of IDH1 ing changes in development, global transcriptional pro- status in gliomas from grades II to IV and support the grams, metabolism, and responses to hypoxia. proposal that its status be incorporated into the current Several studies suggest that IDH1/2 mutation may WHO histopathological scheme for every glioma analyzed be an early event in IDH1/2 mutant neoplasms. When (Hartmann et al. 2010). Despite their histological similar- patients with diffuse astrocytoma or oligoastrocytoma ities, IDH1 mutant and wild-type glioblastomas are clearly were subjected to serial biopsies, there were seven pa- distinct diseases (Fig. 1), and understanding the biological tients who carried only IDH1 mutations at the first basis behind the differences in their natural histories will biopsy but acquired either TP53 mutation or 1p19q loss surely be a major area of focus in the field. at the second biopsy, suggesting a temporal sequence of mutation acquisition. This possibility is also supported by recent sequence analysis of IDH1 and p53 genes in Tumor biological hallmarks in glioma: invasion, a separate study by Lai et al. (2011). Analysis of a large angiogenesis, and tumor-initiating cells panel of grade II–IV astrocytomas showed a higher pro- Glioma cell invasion pensity for Arg-to-Cys substitutions at position 273 in p53 compared with the high rate of Arg-to-His sub- The ability of glioblastoma cells to invade adjacent brain stitutions at position 132 in IDH1, which would be tissue contributes to the major clinical problem in achiev- consistent with a strand asymmetry mechanism (Rodin ing disease control. The majority of glioblastomas treated and Rodin 1998) in which C / T mutations took place initially with standard therapy will recur within several on the transcribed strand in IDH1 but on the nontran- centimeters of the initial tumor location (Hochberg and scribed strand in p53. The result of this mutational Pruitt 1980; Chamberlain 2011). In addition, 10%–20% of asymmetry would be that mutant IDH1 could be expressed patients may harbor ‘‘macroscopic’’ evidence of invasion at in a nonreplicating clone, whereas mutant p53 could be the time of presentation, including multifocal disease— expressed only after DNA replication in S phase. Together, which can even be bihemispheric in the so-called ‘‘butter- these studies provide potential insights into the evolution of fly’’ pattern—at noncontiguous sites in the brain or in- IDH1 mutant cancers and highlight the importance of serial volving spread along white matter tracts and through tissue analysis and the need for careful clonal analysis to subependymal and/or subarachnoid spaces (Parsa et al. fully clarify how these cancers progress. 2005; Chamberlain 2011). It is well established that histologically identifiable tumors invariably greatly ex- Translational relevance of IDH1 status Despite our ceed the area of disease detected radiographically (Burger incomplete understanding of mutant IDH biology, the et al. 1988). The intrinsic capacity of glioblastoma to mutant status of the IDH1/2 genes may serve as an invade was well illustrated by classic autopsy studies, important prognostic indicator. Specifically, patients with which demonstrated that 25%–50% of untreated glioma anaplastic astrocytoma (Parsons et al. 2008; Sanson et al. patients examined harbored histological evidence of 2009; Yan et al. 2009; Hartmann et al. 2010) and glioblas- bilateral disease (Scherer 1940; Matsukado et al. 1961). toma (Yan et al. 2009) harboring mutant IDH1 demon- Notably, even grade II, or ‘‘diffuse,’’ gliomas exhibit strate a significantly longer overall survival compared with invasive capacity such that cells derived from this tumor wild-type IDH1 counterparts and are younger at presenta- grade can be observed histologically at least 2 cm from tion, and this survival benefit has also been observed the main tumor mass when ‘‘supratotal’’ surgical re- in grade II gliomas (Sanson et al. 2009). Patients with sections have been attempted (Yordanova et al. 2011). G-CIMP+ tumors also experience a similar survival benefit Thus, glioma cell invasiveness is a clinical problem in (Noushmehr et al. 2010). In addition, a comprehensive the majority of diffuse gliomas. genomic and clinical analysis of glioblastomas harboring The process of glioblastoma invasion likely involves mutant and wild-type IDH1 suggests that, while histo- sequential adhesion to the (ECM), pathologically similar, these tumors may represent disease degradation of the ECM, and altered cell contractility processes far more disparate than has been appreciated. (Giese and Westphal 1996; Nakada et al. 2007; Tate and Specifically, IDH1 mutant tumors display less contrast Aghi 2009; Onishi et al. 2011). Unlike non-CNS cancers, enhancement, less peritumoral edema, larger initial size, clinically manifest hematogenous or lymphatic spread of greater cystic components, and a greater likelihood of glioblastoma outside the brain is exceedingly rare (Hoffman frontal lobe involvement compared with wild-type tumors and Duffner 1985). Also, the ECM composition and dis- (Lai et al. 2011). In addition, several methods have been tribution in the brain is distinct from other extra-CNS developed to assess IDH1 mutant protein status (Capper tissue sites. A typical collagen-rich basement membrane

GENES & DEVELOPMENT 767 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al.

ECM exists in the glia externa limitans, which covers the ing protein 2 (IGFBP2) (Wang et al. 2003; Levitt et al. cortical surface, and also surrounding cerebral blood 2005) and the forkhead transcription factor FoxM1B. vessels (Rutka et al. 1988; Louis 2006; Gritsenko et al. FoxM1B not only up-regulated MMP-2 and was overex- 2012). However, the brain parenchyma harbors a unique pressed in human glioblastoma specimens (Dai et al. 2007), ECM structure, the perineuronal network, which is but also transformed engineered, immortalized normal a meshwork composed predominantly of hyaluronan human astrocyte cells into invasive glioblastoma cells sulfate proteoglycans as well as chondroitin sulfates pro- via a pathway involving PTEN degradation and AKT teoglycans, tenascins, and link proteins (Kwok et al. 2011; activation (Dai et al. 2010). Notably, AKT also regulates Gritsenko et al. 2012). Moreover, distinct brain regions, the -binding protein Girdin, which directs neural such as the neurogenic areas of the SVZs, are more cell migration and contributes to tumor-initiating cell enriched for chondroitin and heparan sulfate proteogly- invasiveness (Natsume et al. 2011); because several cans (Sirko et al. 2007). invasive mechanisms can be regulated by the PTEN/ Glioma cells adhere to the ECM using several mecha- PI3K/AKT pathway, its contribution to this phenotype nisms. The immunoglobulin superfamily member CD44 deserves further investigation. Given that the MMP and the hyaluronan-mediated motility receptor (RHAMM), inhibitor marimistat did not show efficacy in a ran- both receptors for hyaluronan, are expressed in glioblas- domized clinical trial, albeit as a monotherapy with- toma (Akiyama et al. 2001). CD44 is cleaved by both out temozolomide (Levin et al. 2006), it will be impor- ADAM proteases (Murai et al. 2004) and MMP-9 (Chetty tant to further clarify both the critical proteases and et al. 2012) in a process that promotes motility via cyto- their regulatory pathways that promote glioblastoma skeletal reorganization (Murai et al. 2004; Bourguignon invasiveness. 2008); it is likely that myosin II is also important for glioma A growing number of additional molecules have been cell contractility during invasion (Beadle et al. 2008). implicated in invasion and are reviewed elsewhere Strikingly, CD44 and RHAMM are both suppressed by (Nakada et al. 2007; Teodorczyk and Martin-Villalba p53 (Godar et al. 2008; Sohr and Engeland 2008), suggest- 2010). Importantly, it will be critical to test candidate ing that early cellular progression through canonical check- proinvasive targets and programs in physiologically rele- points and the ability to migrate/invade are linked, vant tumor models with aggressive growth behavior. although this possibility remains to be validated in glioma Conventional malignant glioma cell lines do not tend to models. , particularly the avb3andavb5hetero- exhibit invasive growth in orthotopic xenograft models, dimers (Bello et al. 2001a), also likely contribute to glioma but tumor-initiating cells not only invade (Singh et al. cell adherence to the ECM in a process that activates 2004; Wong et al. 2011), but also often phenocopy the type cytoskeletal rearrangement through cytoplasmic medi- of invasion observed radiographically in patients from ators, including FAK (Rutka et al. 1999; Riemenschneider which the cells were derived (Wakimoto et al. 2012), et al. 2005) and/or Pyk2 (Lipinski et al. 2008). Their thereby representing high-fidelity platforms to explore putative involvement in glioma pathobiology (D’Abaco invasion (Bhat et al. 2011). Furthermore, the striking, but and Kaye 2007; Desgrosellier and Cheresh 2010) has poorly understood, migratory capacity of neural and prompted the testing of the avb3andavb5 mesenchymal stem cells in transplant models (Carney inhibitor cilengitide in an ongoing phase III CENTRIC and Shah 2011) affords another invasive model that can be clinical trial for newly diagnosed glioblastoma in com- used to interrogate this process. bination with radiation therapy and temozolomide (Reardon et al. 2011a). Angiogenesis Several MMPs have been implicated in modifying the ECM in the local microenvironment to promote invasion Angiogenesis plays an important role in glioblastoma (Rao 2003). Recent work revealed a novel mechanism of (Kargiotis et al. 2006), as evidenced by the presence of EGFR-mediated invasion in which EGFR-dependent up- microvascular proliferation (Louis et al. 2007). The recent regulation induced high expression of the IFN-regulated demonstration that therapeutic strategies to inhibit com- factor guanylate-binding protein 1, leading to increased ponents that contribute to angiogenesis have been shown glioma invasion through MMP-1 expression (Li et al. to have some efficacy in glioblastoma (Reardon et al. 2011). MMP-2 and MMP-9 also drive glioma invasion 2011b), including the recent approval of bevacizumab, (Forsyth et al. 1999) and are regulated through several provides the foundation for future studies in this disease. molecular cascades. Both enzymes promoted invasion Glioma cells require blood vessels for metabolic pur- when up-regulated via a CD95-mediated activation of poses, such as oxygen and nutrient delivery and waste AKT1 involving recruitment of Src family member Yes removal, and also for the creation of a vascular niche that and p85. Importantly, CD95L expression was demon- may selectively support glioma stem cells (Calabrese strated at the leading glioma edge in clinical samples et al. 2007; Gilbertson and Rich 2007). The development (Kleber et al. 2008). MMP-2 and MMP-9 were also of glioma vasculature may occur through several mech- coordinately up-regulated by the low-density lipoprotein anisms (Carmeliet and Jain 2011a): angiogenesis, the receptor-related protein 1 in an ERK-dependent promi- formation of new blood vessels from the existing vascu- gratory process (Song et al. 2009). Further work has lature (Folkman 1971; Kerbel 2008); vasculogenesis, revealed additional regulators of MMP-2 expression, in- which involves the recruitment of bone marrow-derived cluding the PTEN-regulated insulin growth factor-bind- endothelial progenitor cells (Patenaude et al. 2010); re-

768 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma cruitment of tumor cells directly into the vascular wall; glioblastoma (Wen et al. 2010). However, it is unclear or the differentiation of tumor stem cells directly into whether these changes result from ‘‘vascular normaliza- vascular endothelium (Ricci-Vitiani et al. 2010; Wang tion’’ (Jain 2005; Carmeliet and Jain 2011b), actions on et al. 2010). The end result of these blood vessel-pro- tumor cells, or other alterations in the blood–brain barrier ducing processes is an intratumoral vasculature that is (Bechmann et al. 2007). In addition, patients on these highly aberrant, incomplete, and tortuous (Long 1970), agents tend to progress rapidly when disease recurs after creating areas of hypoxia, acidosis, and peritumoral treatment with anti-angiogenic agents (Ellis and Hicklin edema. 2008), with minimal response to subsequent chemother- Blood vessel formation is regulated by a balance be- apy (Quant et al. 2009), underscoring the importance of tween pro- and anti-angiogenic molecules that comprise understanding the basis for treatment resistance. Due an angiogenic switch (Bergers and Benjamin 2003). VEGF, to compensatory up-regulation of alternative angiogenic acting through VEGFR-2/KDR, is believed to be the (Batchelor et al. 2007; Sathornsumetee and Rich 2007) or central proangiogenic factor and is induced by hypoxia vasculogenic (Du et al. 2008) pathways, combinations of via HIF-1a (Shweiki et al. 1992; Kaur et al. 2005) and inhibitors that target nonredundant vascular pathways several mitogenic pathways that are dysregulated in may be necessary. In addition, there is some evidence glioblastoma (Maity et al. 2000; Pore et al. 2003; Watnick that angiogenesis inhibitors may promote infiltrative et al. 2003; Phung et al. 2006). VEGFR-2 activation then glioma growth (Norden et al. 2008b; Iwamoto et al. 2009; regulates endothelial cell survival, proliferation, migra- Narayana et al. 2012). An improved understanding of the tion, and permeability (Hicklin and Ellis 2005). In addi- basic biology of angiogenesis, the mechanisms of resis- tion to VEGF, there are a large number of other factors tance to inhibitors, and the contributions of tumor-initi- that stimulate angiogenesis in glioblastoma (Carmeliet ating cell-derived vasculature to the microenvironment and Jain 2011b), including PDGF, FGF, the ANG/TIE will be necessary to optimize angiogenesis inhibition as system, Notch signaling, integrins, ephrins, IL-8, and a therapeutic approach in glioblastoma. SDF-1a (Carmeliet and Jain 2011a; Weis and Cheresh 2011). These proangiogenic mediators are opposed by Glioblastoma ontogeny: cellular origins anti-angiogenic factors, including angiostatin, thrombo- and tumor-initiating cells spondins, endostatin, , and interferons (Nyberg et al. 2005). When stimulatory factors outweigh inhibitory Cellular origins The cellular origins of malignant glio- factors, the angiogenic switch favors blood vessel creation. mas continue to be a source of debate. As in other cancers, Several types of angiogenesis inhibitors have been the continued interest in glioma ontogeny is stimulated by developed for therapeutic use. Because it is the main the possibility that an improved understanding of the driver of angiogenesis, most approaches target VEGF normal cell of origin will help identify fundamental signaling by interfering with either the VEGF ligand pathways and lineage dependencies that could represent (Vredenburgh et al. 2007b), its receptor (VEGFR-2/KDR) novel diagnostic and therapeutic targets (Visvader 2011). (Batchelor et al. 2010), or its downstream signaling cas- Specifically, numerous studies in genetically engineered cade. The anti-VEGF antibody bevacizumab is now used mouse models have provided evidence that gliomas arise heavily in the clinic after undergoing accelerated approval from the normal reservoirs of cycling stem and pro- by the FDA for use in recurrent glioblastoma (Cohen genitor cells within the brain, and numerous genetically et al. 2009). Two initial phase II trials of bevacizumab and engineered mouse models have supported this idea in irinotecan demonstrated a 60% radiographic response and that a diverse range of glioma-relevant mutations tar- an apparent doubling of 6-mo progression-free survival geted to neural stem cells in vivo readily produce gliomas (PFS; 38%–46%) and median survival (40–42 wk) com- with high fidelity and penetrance (Bachoo et al. 2002; Zhu pared with historical controls (9%–15% and 22–26 wk, et al. 2005; Zheng et al. 2008). For instance, tamoxifen- respectively) (Vredenburgh et al. 2007a,b). In addition, inducible Cre-recombinase-mediated inactivation of p53 two follow-up phase II trials demonstrated a radiographic and NF1 in adult stem cells drove the consistent formation response (27%–38%), increased PFS (29%–50%), and a of astrocytomas (Alcantara Llaguno et al. 2009; Wang et al. slightly prolonged overall survival (31–35 wk) compared 2009); interestingly, these tumors were not restricted to with historic controls (Friedman et al. 2009; Kreisl et al. the SVZ neural stem cell niche location but were found 2009); phase III trials, including the RTOG-0825 study, within multiple other brain regions. This finding was are currently under way (http://www.clinicaltrials.gov). extended by C Liu et al. (2011) using the mosaic analysis Additional approaches, such as angiopoietin inhibitors, with double markers (MADM) technique (Zong et al. are reviewed elsewhere (Norden et al. 2008a; Reardon 2005). This approach enabled careful longitudinal, lineage et al. 2011b). tracing analysis of developing astrocytomas in mice with However, several unanswered clinical questions re- mosaic p53/NF1 mutant neural stem cells, which showed main that require an improved understanding of the basic that the accelerated phase of tumor growth occurs not in mechanisms involved in glioma angiogenesis (Verhoeff the original cell of mutation, but within OLIG2+ oligo- et al. 2009). For example, anti-angiogenic agents reduce dendroglial progenitor cells (OPCs) that migrate away vasogenic edema and corticosteroid requirements by de- from the niche zones (C Liu et al. 2011). The finding that creasing vascular permeability (Gerstner et al. 2009), ‘‘astrocytomas’’ can arise from OPCs in mice could help thereby dramatically modifying the MRI appearance of to explain the long-standing observation that human

GENES & DEVELOPMENT 769 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al. glioblastoma and all other astrocytomas are characterized ences in location between IDH1 mutant and wild-type by both expression of oligodendroglial markers to an even glioblastomas (Lai et al. 2011). Given the diverse number greater degree than astrocytic markers (Ligon et al. 2004; of histological subtypes, various subsets of molecular Phillips et al. 2006; Verhaak et al. 2010) and their de- patterns and subclasses, and increasingly broad number pendency on the same lineage factors (e.g., OLIG2) as of stem and progenitor cells in the brain, further studies normal counterparts (Ligon et al. 2007). Distributed pro- will be required to clarify the origins of malignant genitor cells, such as OPCs, actually represent the largest gliomas and the programs that drive their subsequent pool of cycling cells in the brain and are defined by evolution. expression of OLIG2 and NG2 (Dawson et al. 2003; Geha et al. 2010). Targeting of these nonstem cell progenitors Tumor-initiating cells Complementing these studies of through transgenic or viral approaches has been shown to the cell of origin in glioblastoma, several groups have lead to malignant astrocytomas or oligodendrogliomas, shown that different cell populations within a tumor depending on the combination of mutations and cell exhibit varied tumor-forming capacity. These observa- types targeted (Geha et al. 2010; Persson et al. 2010). tions suggest that stem/progenitor cells may serve as the In addition, work in genetically engineered mice has cell of origin for many tumors. Irrespective of the ontog- demonstrated that gliomas may also arise from terminally eny argument, however, the finding that distinct cellular differentiated cells, likely through a process of dedifferen- subpopulations within glioblastomas harbor potent tu- tiation. Bachoo et al. (2002) demonstrated that cultured mor-initiating capacity was yet another example that mature Ink4a/ArfÀ astrocytes stimulated with EGF or heterogeneity—morphological, molecular, and cellular— expressing mutant EGFRvIII dedifferentiated to nestin- is the rule, rather than the exception, in this group of expressing progenitor cells capable of forming high-grade diseases. First described by several groups (Ignatova et al. gliomas in vivo, providing evidence that both progenitor 2002; Hemmati et al. 2003; Singh et al. 2003), the finding and lineage-restricted, mature cells were permissive con- that tumor-initiating cancer cells can be isolated from texts for glioma formation. Moreover, PDGF has also glioblastomas and other cancers and propagated as neuro- been shown to induce the dedifferentiation of mature spheres when grown in serum-free neural stem cell astrocytes and induce glioma formation in GFAP-express- conditions, have self-renewal capacity, and recapitulate ing Ink4a/ArfÀ cells in vivo (Dai et al. 2001). Together, malignant glioma behavior in vivo (Singh et al. 2004) has these studies emphasize the influence of particular com- transformed glioblastoma research. Moreover, these cells binations of dysregulated pathways on glioma develop- in malignant gliomas now represent well-characterized ment and highlight the notion that specific genetic alter- experimental models for the investigation of the cancer ations, and not just the precise a priori developmental state stem cell hypothesis (Dirks 2010), since these cells of the cell, help to create permissive contexts for glioma represent highly faithful glioma models when cultured formation. as tumor-initiating cell lines or orthotopic xenografts. For Although murine studies have been helpful in clarify- instance, culturing tumor specimens in stem cell condi- ing glioma origins, it remains to be seen whether their tions has enabled the maintenance of lines carrying glioma- findings can be fully extrapolated to human disease. For specific lesions such as EGFRvIII (Stockhausen et al. 2011) instance, human and mouse OPCs may be biologically or mutant IDH1 expression (Luchman et al. 2012) that similar, as recent work demonstrated conserved mecha- were not well-modeled under serum-containing condi- nisms for oligodendroglioma formation through disrup- tions. In addition, abundant evidence confirms that these tion of asymmetric division of NG2+ OPCs in mouse models recapitulate the phenotypes as well as genomic verbB/p53+/À-induced oligodendrogliomas and human alterations found in patient tumors (Lee et al. 2006a; oligodendrogliomas (Sugiarto et al. 2011). However, in Wakimoto et al. 2012). The glioma stem cell hypothesis adult mice, the SVZ is a prominent regenerative zone, but (Dirks 2010; Taylor et al. 2010; Lathia et al. 2011a) the analogous region in adult human brains, the subven- incorporates a model in which these tumor-initiating cells tricular astrocyte ribbon, may not harbor similar num- are the principal drivers of gliomagenesis. Specifically, bers or types of stem cells (Sanai et al. 2011). In addition, these cells are thought to represent transformed neural the plasticity of cells in response to engineered oncogene progenitors that give rise to tumor cell progeny and also and tumor suppressor alterations is well established and have the capacity to self-renew. Considering the high thus can confound deconvolution of tumor-initiating malignant potential of the tumor-initiating cell popula- populations from established tumors. For instance, stud- tion, this hierarchical model implies that this subset, ies of the effects of Smo activation on development of which may represent a minority of cells within the bulk a mouse model of medulloblastoma suggested that the tumor mass, is the critical subpopulation of glioblas- same mutation at multiple stages of lineage commitment toma that could be highly relevant clinically due to the could result in dedifferentiation of committed cells to potential for intrinsic therapeutic resistance (Bao et al. a common phenotypic end point, thereby obscuring the 2006). In contrast, an alternative view of gliomagenesis original origin of the tumors (Schuller et al. 2008). It will involves a less hierarchical model in which most glio- be important to continue to model human disease in mice blastoma cells are functionally and developmentally sim- given the powerful technologies enabling sophisticated ilar and have developed through a progressive process of genetic targeting. These approaches will be invaluable to clonal selection of transformed somatic cells. In this case, explore clinical observations such as the distinct differ- there may be a higher proportion of the tumor mass with

770 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma malignant potential than in the stem cell model. Further human—that recapitulate the salient in vivo behaviors work will be important to discern the physiological of interest. In addition to survival, phenotypes such as relevance and potential translational impact of these invasion should also be assessed. Systematic approaches not mutually exclusive models. may also be useful in revealing synthetic-lethal relation- Although the tumor-initiating cell subpopulation has ships (Barbie et al. 2009), such as the role of ENTPD5 in been traditionally defined by the cell surface expression the context of PTEN loss (Fang et al. 2010). In addition to of CD133, it has become clear that this marker does not systematic approaches, the computational methodolo- exclusively define tumor-initiating populations (Ogden gies involving gene expression data that have pointed to et al. 2008; Chen et al. 2010; Beier and Beier 2011), and important biological roles for STAT3 (Carro et al. 2010; thus further work correlating cell surface markers (Lathia Bhat et al. 2011) and the Hippo pathways, among others, et al. 2011b) with phenotype is necessary to more clearly have shown great promise and should be explored define these cells. Moreover, while substantial work has further in preclinical contexts. begun to clarify the programs—including PLAGL2, Olig2, and c-Myc (Ligon et al. 2007; Zheng et al. 2008, 2010; Understanding malignant transformation Guryanova et al. 2011)—that drive and maintain tumor- The phenomenon of transformation from grade II glioma initiating characteristics, additional study is needed re- to malignant disease is poorly understood yet accounts garding the molecular pathways underlying cellular phe- for the mortality of low-grade disease. These patients notype and hierarchies. Also, the degree to which these are younger at diagnosis, and it is thought that 45%– cells strictly preserve the phenotypes of all genotypes of 70% will undergo disease progression to histological ma- original patient tumors remains to be examined in large lignancy within 5 years (Ohgaki and Kleihues 2005). numbers of patients in detail; reports of variation with Although the entity of secondary glioblastoma repre- respect to preservation of RTK, bias toward tumors with sents <10% of all diagnosed glioblastomas (Ohgaki and PTEN inactivation, and difficulties in modeling the Kleihues 2009), the existence of histologically proven vascular changes seen in human glioblastoma suggest antecedent lower-grade tissue presents a unique and as yet that careful attention to the unique properties of each line relatively underexploited opportunity to understand the will be required (Chen et al. 2010). Nevertheless, the use programs that drive malignancy and also to explore the of glioblastoma tumor-initiating cell models in the basic possibility of delaying clinical transformation. The study research setting is now routine, and numerous emerging of cancers that progress from lower-grade lesions to higher- preclinical studies using patient-derived models will pro- grade malignancies has provided critical insights into the vide clues as to whether these lines gathered into patient genetic basis of cancer evolution. Specifically, analysis of cohorts might advance drug screening efforts (Pollard the molecular changes along the histopathological spec- et al. 2009). trum in colorectal cancer (Vogelstein et al. 1988) pioneered comparative analyses of lower-grade lesions and their Future directions: areas of further study higher-grade counterparts. Together, the dissection of the chronological genomic events that drive disease progres- Identifying dependencies and targets sion not only provides invaluable fundamental mechanis- tic information on a cancer’s natural history, but also may Despite the tremendous progress in our understanding of identify potential therapeutic targets that drive the tran- the genetic basis of glioblastoma, targeted therapeutic sition from less aggressive to more malignant disease. approaches based on known genomic alterations have not Especially considering the comparatively young age of proved to be efficacious to date (De Witt Hamer 2010). It presentation in low-grade patients, the benefit to inter- is likely that intratumoral heterogeneity and target coop- rupting or delaying the transformation pathway would be erativity (Stommel et al. 2007) conspire to create a ‘‘mul- significant. Moreover, despite the clear differences be- tiple dependency’’ state wherein single-target inhibitors tween IDH1 mutant secondary glioblastoma and IDH1 are not sufficient to attenuate tumor growth. Moreover, wild-type primary disease, the isolation of transformation transcriptionally defined subclasses may harbor unique programs identified by studying low-grade and matched dependencies. In addition, we have an incomplete un- malignant disease could be relevant to the programs, such derstanding of the functional consequences of many of as those driving angiogenesis, that contribute to the bi- the genes found mutated in glioblastoma. Indeed, it will ology of primary disease. Earlier studies comparing pri- be necessary to approach functional space in glioblas- mary and secondary glioblastomas (Maher et al. 2006; Tso toma in the systematic manner with which the struc- et al. 2006), characterizing the expression of DCC in low- tural space has been clarified (Boehm and Hahn 2011). To and high-grade tumors (Reyes-Mugica et al. 1997), and this end, genome-wide tools capable of loss- and gain-of- performing comparative array CGH between a small set of function studies will likely be invaluable in interrogat- grade II/IV gliomas (Idbaih et al. 2008) point to the ing phenotypes particularly relevant to glioblastoma. differences in these tumors. Large-scale pooled shRNA screening of a panel of glio- blastoma cell lines as part of a larger panel of multiple Understanding the extent and impact of heterogeneity lineages led to the identification of a list of candidate dependencies (Cheung et al. 2011). These approaches The morphological heterogeneity that prompted the origi- should use panels of highly faithful models—murine and nal description of high-grade glioma as ‘‘glioblastoma multi-

GENES & DEVELOPMENT 771 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al. forme’’ (Mallory 1914; Bailey and Cushing 1926) has also both intrinsic and acquired mechanisms of resistance extended to the molecular level. Through studies demon- play significant roles in glioblastoma. Currently, most strating variable intratumoral distributions of EGFRvIII- patients continue to receive temozolomide as the main expressing cells (Inda et al. 2010) as well as mutually chemotherapeutic agent in their treatment regimen. In exclusive SCNAs within distinct populations of the addition to the de novo resistance mechanism of MGMT same tumor (Snuderl et al. 2011; Szerlip et al. 2012), hypomethlyation (Hegi et al. 2005), several groups have among others, we are beginning to have a better un- described additional mechanisms of treatment resistance derstanding of how molecular heterogeneity is manifest. that include loss of the mismatch repair gene MSH6 While these studies illustrate examples of distinct alter- (Cahill et al. 2007), silencing of the base excision repair ations occurring in unique cells, there are also examples enzyme alkylpurine-DNA-N-glycosylase (Agnihotri et al. of concomitant kinase activation within the same cell 2012), and also overexpression of both inhibitor of apo- (Stommel et al. 2007). In both cases, it is not entirely ptosis proteins (Ziegler et al. 2008) and multidrug re- clear whether intermingled cell populations with unique sistance members (Shervington and Lu genomic alterations behave as independent tumors or are 2008). Although perturbation of recurrent resistance mech- interdependent with each other. Ultimately, it will be anisms could enhance the therapeutic efficacy of this necessary to study the populations of a heterogeneous current workhorse agent, these studies point to the tumor mass at the single-cell level in order to under- broader pivotal framework of carefully studying pre- stand co-occurring genetic alterations and accompany- and post-intervention tumor samples, as has been dem- ing protein activation states. Developing technology onstrated in other cancers. Comparative post-treatment capable of assessing single-cell-scale genomes, isolated annotation will be important for additional agents, such through laser capture microscopy or live cell-sorting as bevacizumab, and perhaps every emerging targeted techniques, is making these approaches feasible (Navin agent that enters clinical trials. This approach will also and Hicks 2011; Navin et al. 2011). Moreover, in addition facilitate our understanding of acquired resistance and/ to understanding the molecular basis of tumor cell hetero- or de novo insensitivity (Bao et al. 2006) to radiation geneity, it will be critical to consider the contributions and therapy, as the vast majority of post-treatment tissue has phenotypes of the diverse nontumor cell types, such as been subjected to up-front radiation treatment. We antici- stromal and immune/inflammatory cells (Dunn et al. pate that components of the tumor microenvironment may 2007), that populate the glioma microenvironment. also contribute to resistance in a tumor cell-extrinsic manner. Specifically, the blood–brain barrier, formed by Developmental oncobiology intracerebral capillary tight junctions and astrocytic foot processes, impairs the passive transit of molecules between Given the growing prominence of stem and progenitor the vascular lumen and brain parenchyma (Huber et al. 2001) cell biology in glioblastoma, further characterization of and is highly dysregulated in glioblastoma (Long 1970). This the heterogeneous cell populations found within these structure is known to present a challenge to effective transit tumors will be critical in understanding their origins and of therapeutic molecules to the brain, and thus continued evolution. Although it is clear that defects in develop- study of its biology in both the normal and tumor settings mental hierarchy play a key role in many tumors, further could enhance the efficacy of drug delivery to glioblastoma. work is necessary to determine the molecular basis un- derlying the differentiation potential of the cells within each tumor. Specifically, comprehensive genomic, epige- Toward a molecular classification of glioma nomic, and transcriptional characterization of carefully The comprehensive molecular characterization of glio- isolated progenitor cells from tumors compared with their mas is now starting to transform their classification nonneoplastic counterparts may clarify the structural (Table 1), which currently follows the consensus WHO basis of particular cell states, whether there are distinct histopathological criteria instrumental in standardizing differences across the transcriptomic subtypes in glio- pathological classification of these challenging tumors blastoma, and how the neoplastic context influences (Louis et al. 2007). However, a high percentage of gliomas, phenotypes such as invasion and migration, resistance such as mixed oligoastrocytomas and lower-grade glio- to therapy, and signaling dysregulation. Furthermore, these mas, remain difficult to categorize reproducibly due to types of comparative studies may enable the development considerable histological overlap. Genomic approaches of therapeutic approaches that reduce the tumor-initiating/ applied to clinically characterized patient cohorts now tumor-propagating capacities of glioma tumor-initiating clearly show that combined molecular and histological cells by interrupting the cellular programs that likely classification offers great opportunities to significantly contribute to their aggressiveness, such as self-renewal, improve clinical predictive power over use of histology maintenance of multipotentiality, and the phenotypic alone, even for individual patients. A clear example is the plasticity to dedifferentiate. group including oligodendrogliomas, grade II and III astrocytomas, and mixed gliomas, which are best defined Therapeutic resistance as a subset of gliomas harboring shared genomic features Given the inevitable progression of glioblastoma follow- of mutations in IDH genes and are G-CIMP+ and MGMT ing standard-of-care surgery, temozolomide treatment, methylated due to globally increased levels of methyla- and radiation therapy (Stupp et al. 2005), it is likely that tion (Noushmehr et al. 2010).

772 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma

Further subclassification of this group will likely use (to L.C. and R.A.D.); NIH Grant K08-NS062907 (to M.G.C.); an unique lesions in oligodendroglioma, such as 1p/19q award from the Sontag Foundation (to K.L.); an ABTA Basic whole-arm codeletion/translocation, CIC mutations, and Research Fellowship (to G.P.D.); NINDS R25 award (to G.P.D.); 1p or 19q polysomy (Snuderl et al. 2009; Bettegowda et al. and the Brain Science Foundation (to I.F.D.). W.K.C. is a fellow of 2011; Yip et al. 2012), in the presence or absence of TP53 the National Foundation for Cancer Research. mutations to enable reliable identification of a group in a manner not possible by histology alone. Likewise, References BRAF fusions (Horbinski et al. 2010; Hawkins et al. Agnihotri S, Gajadhar AS, Ternamian C, Gorlia T, Diefes KL, 2011) and V600E mutants characterize additional astro- Mischel PS, Kelly J, McGown G, Thorncroft M, Carlson BL, cytoma subsets with clinical relevance (Dias-Santagata et al. 2012. Alkylpurine-DNA-N-glycosylase confers resis- et al. 2011; Schindler et al. 2011; Tian et al. 2011). In tance to temozolomide in xenograft models of glioblastoma contrast to the mutant IDH/G-CIMP+ tumors, glioblas- multiforme and is associated with poor survival in patients. tomas appear surprisingly distinct but significantly more J Clin Invest 122: 253–266. complex genomically, making diagnostic subclassifica- Akhavan D, Cloughesy TF, Mischel PS. 2010. mTOR signaling in glioblastoma: Lessons learned from bench to bedside. tion schemes more challenging. Deeper sequencing ef- Neuro Oncol 12: 882–889. forts have yet to be fully completed and may yet reveal Akiyama Y, Jung S, Salhia B, Lee S, Hubbard S, Taylor M, novel recurrent mutations that draw together what ap- Mainprize T, Akaishi K, van Furth W, Rutka JT. 2001. pears now to be a highly diverse group of tumors. Clearly, Hyaluronate receptors mediating glioma cell migration and mutant IDH will be critical in discerning these diseases, proliferation. J Neurooncol 53: 115–127. and translation of expression profiling work into diagnos- Alcantara Llaguno S, Chen J, Kwon CH, Jackson EL, Li Y, Burns tic tools is also under way (Colman et al. 2010). Additional DK, Alvarez-Buylla A, Parada LF. 2009. Malignant astrocyto- diagnostic scenarios, such as distinction of pseudoprogres- mas originate from neural stem/progenitor cells in a somatic sion versus true tumor progression, will hopefully benefit tumor suppressor mouse model. Cancer Cell 15: 45–56. from molecular pathology approaches. The rapid pace of Amary MF, Bacsi K, Maggiani F, Damato S, Halai D, Berisha F, Pollock R, O’Donnell P, Grigoriadis A, Diss T, et al. 2011a. genomic and epigenomic discovery in gliomas will likely IDH1 and IDH2 mutations are frequent events in central necessitate the development of consensus mechanisms to chondrosarcoma and central and periosteal chondromas but deploy integrated molecular and histologic data to clinical not in other mesenchymal tumours. J Pathol 224: 334–343. practice to improve diagnosis and predictive capability and Amary MF, Damato S, Halai D, Eskandarpour M, Berisha F, stratify response to therapy. Bonar F, McCarthy S, Fantin VR, Straley KS, Lobo S, et al. 2011b. Ollier disease and Maffucci syndrome are caused by somatic mosaic mutations of IDH1 and IDH2. Nat Genet 43: Conclusion 1262–1265. Assanah M, Lochhead R, Ogden A, Bruce J, Goldman J, Canoll P. The commitment to genomic characterization of glio- 2006. Glial progenitors in adult white matter are driven to blastoma has fueled substantial progress in our under- form malignant gliomas by platelet-derived growth factor- standing of this cancer in the last 5 years. The heteroge- expressing retroviruses. J Neurosci 26: 6781–6790. neous nature of this neoplasm, both within and across Assanah MC, Bruce JN, Suzuki SO, Chen A, Goldman JE, Canoll tumors, underscores the difficulty in developing effica- P. 2009. PDGF stimulates the massive expansion of glial cious treatments and provides a challenge both to anno- progenitors in the neonatal forebrain. Glia 57: 1835–1847. tate tumors to understand the disease and to stratify Bachoo RM, Maher EA, Ligon KL, Sharpless NE, Chan SS, You MJ, Tang Y, DeFrances J, Stover E, Weissleder R, et al. 2002. patients for trials and treatment. Already there are several Epidermal growth factor receptor and Ink4a/Arf: Convergent efforts to connect the information obtained from retro- mechanisms governing terminal differentiation and trans- spective, population-based genomic characterization to formation along the neural stem cell to astrocyte axis. the individual patient in the form of multiplexed patient Cancer Cell 1: 269–277. profiling on clinical samples (MacConaill et al. 2009; Bailey P, Cushing H. 1926. A classification of the tumors of the Sequist et al. 2011). Further work will be necessary to glioma group on a histogenetic basis with a correlated study fully understand glioblastoma biology and will increas- of prognosis. Lippincott, Philadelphia. ingly require integrated studies, including genomics, Bajenaru ML, Zhu Y, Hedrick NM, Donahoe J, Parada LF, faithful animal models of disease (Lim et al. 2011; Wee Gutmann DH. 2002. Astrocyte-specific inactivation of the et al. 2012), and the careful study of human tissue. neurofibromatosis 1 gene (NF1) is insufficient for astrocy- toma formation. Mol Cell Biol 22: 5100–5113. Bajenaru ML, Hernandez MR, Perry A, Zhu Y, Parada LF, Acknowledgments Garbow JR, Gutmann DH. 2003. Optic nerve glioma in mice requires astrocyte Nf1 gene inactivation and Nf1 brain We thank Tanaz Sharifinia and Ravindra Uppaluri for comments heterozygosity. Cancer Res 63: 8573–8577. on the manuscript. This work was supported by NIH Grants PO1 Balss J, Meyer J, Mueller W, Korshunov A, Hartmann C, von CA117969, RC2 CA148268, and PO1 CA142536 (to W.C.H., Deimling A. 2008. Analysis of the IDH1 codon 132 mutation K.L., L.C., and R.A.D.); the Ben and Catherine Ivy Foundation (to in brain tumors. Acta Neuropathol 116: 597–602. L.C., R.A.D., and K.L.); an award from the Goldhirsh Foundation Banerjee S, Byrd JN, Gianino SM, Harpstrite SE, Rodriguez FJ, (to F.F. and K.L.); NIH (NINDS) Fellowship Award F32 NS066519 Tuskan RG, Reilly KM, Piwnica-Worms DR, Gutmann DH. (to J.W.); NIH Grant P01 CA95616 (to C.B, K.L., L.C., W.C.H., 2010. The neurofibromatosis type 1 tumor suppressor con- W.K.C., F.F., and R.A.D.); NIH Grant NCI-MMHCC-CA84313-06 trols cell growth by regulating signal transducer and activa-

GENES & DEVELOPMENT 773 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al.

tor of transcription-3 activity in vitro and in vivo. Cancer mesenchymal differentiation in malignant glioma. Genes Res 70: 1356–1366. Dev 25: 2594–2609. Banerjee S, Crouse NR, Emnett RJ, Gianino SM, Gutmann DH. Biernat W, Huang H, Yokoo H, Kleihues P, Ohgaki H. 2004. 2011. Neurofibromatosis-1 regulates mTOR-mediated astro- Predominant expression of mutant EGFR (EGFRvIII) is rare cyte growth and glioma formation in a TSC/Rheb-independent in primary glioblastomas. Brain Pathol 14: 131–136. manner. Proc Natl Acad Sci 108: 15996–16001. Bleau AM, Hambardzumyan D, Ozawa T, Fomchenko EI, Huse Bao S, Wu Q, McLendon RE, Hao Y, Shi Q, Hjelmeland AB, JT, Brennan CW, Holland EC. 2009. PTEN/PI3K/Akt pathway Dewhirst MW, Bigner DD, Rich JN. 2006. Glioma stem cells regulates the side population phenotype and ABCG2 activity promote radioresistance by preferential activation of the in glioma tumor stem-like cells. Cell Stem Cell 4: 226–235. DNA damage response. Nature 444: 756–760. Boehm JS, Hahn WC. 2011. Towards systematic functional Bar EE, Lin A, Tihan T, Burger PC, Eberhart CG. 2008. Frequent characterization of cancer genomes. Nat Rev Genet 12: gains at chromosome 7q34 involving BRAF in pilocytic 487–498. astrocytoma. J Neuropathol Exp Neurol 67: 878–887. Boerner JL, Demory ML, Silva C, Parsons SJ. 2004. Phosphory- Barbie DA, Tamayo P, Boehm JS, Kim SY, Moody SE, Dunn IF, lation of Y845 on the epidermal growth factor receptor Schinzel AC, Sandy P, Meylan E, Scholl C, et al. 2009. mediates binding to the mitochondrial protein cytochrome Systematic RNA interference reveals that oncogenic KRAS- c oxidase subunit II. Mol Cell Biol 24: 7059–7071. driven cancers require TBK1. Nature 462: 108–112. Bonavia R, Inda MM, Vandenberg S, Cheng SY, Nagane M, Basso K, Margolin AA, Stolovitzky G, Klein U, Dalla-Favera R, Hadwiger P, Tan P, Sah DW, Cavenee WK, Furnari FB. 2011. Califano A. 2005. Reverse engineering of regulatory net- EGFRvIII promotes glioma angiogenesis and growth through works in human B cells. Nat Genet 37: 382–390. the NF-kB, interleukin-8 pathway. Oncogene doi: 10.1038/ Batchelor TT, Sorensen AG, di Tomaso E, Zhang WT, Duda DG, onc.2011.563. Cohen KS, Kozak KR, Cahill DP, Chen PJ, Zhu M, et al. 2007. Borger DR, Tanabe KK, Fan KC, Lopez HU, Fantin VR, Straley AZD2171, a pan-VEGF inhibitor, KS, Schenkein DP, Hezel AF, Ancukiewicz M, Liebman HM, normalizes tumor vasculature and alleviates edema in glio- et al. 2012. Frequent mutation of isocitrate dehydrogenase blastoma patients. Cancer Cell 11: 83–95. (IDH)1 and IDH2 in cholangiocarcinoma identified through Batchelor TT, Duda DG, di Tomaso E, Ancukiewicz M, Plotkin broad-based tumor genotyping. Oncologist 17: 72–79. SR, Gerstner E, Eichler AF, Drappatz J, Hochberg FH, Benner Bourguignon LY. 2008. Hyaluronan-mediated CD44 activation T, et al. 2010. Phase II study of cediranib, an oral pan-vascular of RhoGTPase signaling and function promotes endothelial growth factor receptor tyrosine kinase inhibitor, tumor progression. Semin Cancer Biol 18: 251–259. in patients with recurrent glioblastoma. JClinOncol28: Bredel M, Bredel C, Juric D, Harsh GR, Vogel H, Recht LD, Sikic 2817–2823. BI. 2005. High-resolution genome-wide mapping of genetic Beadle C, Assanah MC, Monzo P, Vallee R, Rosenfeld SS, Canoll alterations in human glial brain tumors. Cancer Res 65: P. 2008. The role of myosin II in glioma invasion of the brain. 4088–4096. Mol Biol Cell 19: 3357–3368. Bredel M, Scholtens DM, Harsh GR, Bredel C, Chandler JP, Bechmann I, Galea I, Perry VH. 2007. What is the blood–brain Renfrow JJ, Yadav AK, Vogel H, Scheck AC, Tibshirani R, barrier (not)? Trends Immunol 28: 5–11. et al. 2009. A network model of a cooperative genetic Beier CP, Beier D. 2011. CD133 negative cancer stem cells in landscape in brain tumors. JAMA 302: 261–275. glioblastoma. Front Biosci 3: 701–710. Bredel M, Scholtens DM, Yadav AK, Alvarez AA, Renfrow JJ, Bello L, Francolini M, Marthyn P, Zhang J, Carroll RS, Nikas Chandler JP, Yu IL, Carro MS, Dai F, Tagge MJ, et al. 2011. DC, Strasser JF, Villani R, Cheresh DA, Black PM. 2001a. NFKBIA deletion in glioblastomas. NEnglJMed364: 627–637. avb3 and avb5 integrin expression in glioma periphery. Brennan C. 2011. Genomic profiles of glioma. Curr Neurol Neurosurgery 49: 380–390. Neurosci Rep 11: 291–297. Bello L, Lucini V, Carrabba G, Giussani C, Machluf M, Pluderi Brennan C, Momota H, Hambardzumyan D, Ozawa T, Tandon M, Nikas D, Zhang J, Tomei G, Villani RM, et al. 2001b. A, Pedraza A, Holland E. 2009. Glioblastoma subclasses can Simultaneous inhibition of glioma angiogenesis, cell pro- be defined by activity among pathways liferation, and invasion by a naturally occurring fragment of and associated genomic alterations. PLoS ONE 4: e7752. doi: human metalloproteinase-2. Cancer Res 61: 8730–8736. 10.1371/journal.pone.0007752. Bergers G, Benjamin LE. 2003. Tumorigenesis and the angio- Burger PC, Heinz ER, Shibata T, Kleihues P. 1988. Topographic genic switch. Nat Rev Cancer 3: 401–410. anatomy and CT correlations in the untreated glioblastoma Beroukhim R, Getz G, Nghiemphu L, Barretina J, Hsueh T, multiforme. J Neurosurg 68: 698–704. Linhart D, Vivanco I, Lee JC, Huang JH, Alexander S, et al. Cahill DP, Levine KK, Betensky RA, Codd PJ, Romany CA, 2007. Assessing the significance of chromosomal aberrations Reavie LB, Batchelor TT, Futreal PA, Stratton MR, Curry in cancer: Methodology and application to glioma. Proc Natl WT, et al. 2007. Loss of the mismatch repair protein MSH6 Acad Sci 104: 20007–20012. in human glioblastomas is associated with tumor progres- Beroukhim R, Mermel CH, Porter D, Wei G, Raychaudhuri S, sion during temozolomide treatment. Clin Cancer Res 13: Donovan J, Barretina J, Boehm JS, Dobson J, Urashima M, 2038–2045. et al. 2010. The landscape of somatic copy-number alteration Cairncross JG, Ueki K, Zlatescu MC, Lisle DK, Finkelstein DM, across human cancers. Nature 463: 899–905. Hammond RR, Silver JS, Stark PC, Macdonald DR, Ino Y, Bettegowda C, Agrawal N, Jiao Y, Sausen M, Wood LD, Hruban et al. 1998. Specific genetic predictors of chemotherapeutic RH, Rodriguez FJ, Cahill DP, McLendon R, Riggins G, et al. response and survival in patients with anaplastic oligoden- 2011. Mutations in CIC and FUBP1 contribute to human drogliomas. J Natl Cancer Inst 90: 1473–1479. oligodendroglioma. Science 333: 1453–1455. Calabrese C, Poppleton H, Kocak M, Hogg TL, Fuller C, Hamner Bhat KP, Salazar KL, Balasubramaniyan V, Wani K, Heathcock L, B, Oh EY, Gaber MW, Finklestein D, Allen M, et al. 2007. A Hollingsworth F, James JD, Gumin J, Diefes KL, Kim SH, perivascular niche for brain tumor stem cells. Cancer Cell et al. 2011. The transcriptional coactivator TAZ regulates 11: 69–82.

774 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma

The Cancer Genome Atlas Research Network. 2008. Compre- Christensen JG, Zou HY, Arango ME, Li Q, Lee JH, McDonnell hensive genomic characterization defines human glioblas- SR, Yamazaki S, Alton GR, Mroczkowski B, Los G. 2007. toma genes and core pathways. Nature 455: 1061–1068. Cytoreductive antitumor activity of PF-2341066, a novel Capper D, Zentgraf H, Balss J, Hartmann C, von Deimling A. inhibitor of anaplastic lymphoma kinase and c-Met, in 2009. Monoclonal antibody specific for IDH1 R132H muta- experimental models of anaplastic large-cell lymphoma. tion. Acta Neuropathol 118: 599–601. Mol Cancer Ther 6: 3314–3322. Capper D, Simon M, Langhans CD, Okun JG, Tonn JC, Weller Clarke ID, Dirks PB. 2003. A human brain tumor-derived M, Deimling AV, Hartmann C. 2011. 2-Hydroxyglutarate PDGFR-a deletion mutant is transforming. Oncogene 22: concentration in serum from patients with gliomas does not 722–733. correlate with IDH1/2 mutation status or tumor size. Int J Cloughesy TF, Yoshimoto K, Nghiemphu P, Brown K, Dang J, Cancer doi: 10.1002/ijc.26425. Zhu S, Hsueh T, Chen Y, Wang W, Youngkin D, et al. 2008. Carmeliet P, Jain RK. 2011a. Molecular mechanisms and clin- Antitumor activity of rapamycin in a Phase I trial for patients ical applications of angiogenesis. Nature 473: 298–307. with recurrent PTEN-deficient glioblastoma. PLoS Med 5: e8. Carmeliet P, Jain RK. 2011b. Principles and mechanisms of doi: 10.1371/journal.pmed.0050008. vessel normalization for cancer and other angiogenic dis- Cohen MH, Shen YL, Keegan P, Pazdur R. 2009. FDA drug eases. Nat Rev Drug Discov 10: 417–427. approval summary: Bevacizumab (Avastin) as treatment of Carney BJ, Shah K. 2011. Migration and fate of therapeutic stem recurrent glioblastoma multiforme. Oncologist 14: 1131– cells in different brain disease models. Neuroscience 197: 37– 1138. 47. Colman H, Zhang L, Sulman EP, McDonald JM, Shooshtari NL, Carro MS, Lim WK, Alvarez MJ, Bollo RJ, Zhao X, Snyder EY, Rivera A, Popoff S, Nutt CL, Louis DN, Cairncross JG, et al. Sulman EP, Anne SL, Doetsch F, Colman H, et al. 2010. The 2010. A multigene predictor of outcome in glioblastoma. transcriptional network for mesenchymal transformation of Neuro Oncol 12: 49–57. brain tumours. Nature 463: 318–325. D’Abaco GM, Kaye AH. 2007. Integrins: Molecular determi- CBTRUS. 2011. CBTRUS statistical report: Primary brain and nants of glioma invasion. J Clin Neurosci 14: 1041–1048. central nervous system tumors diagnosed in the United Daginakatte GC, Gutmann DH. 2007. Neurofibromatosis-1 States in 2004–2007. Central Brain Tumor Registry of the (Nf1) heterozygous brain microglia elaborate paracrine fac- United States, Hinsdale, IL. tors that promote Nf1-deficient astrocyte and glioma growth. Chabner BA. 2011. Early accelerated approval for highly targeted Hum Mol Genet 16: 1098–1112. cancer drugs. N Engl J Med 364: 1087–1089. Dai C, Celestino JC, Okada Y, Louis DN, Fuller GN, Holland Chamberlain MC. 2011. Radiographic patterns of relapse in EC. 2001. PDGF autocrine stimulation dedifferentiates cul- glioblastoma. J Neurooncol 101: 319–323. tured astrocytes and induces oligodendrogliomas and oli- Chapman PB, Hauschild A, Robert C, Haanen JB, Ascierto P, goastrocytomas from neural progenitors and astrocytes in Larkin J, Dummer R, Garbe C, Testori A, Maio M, et al. vivo. Genes Dev 15: 1913–1925. 2011. Improved survival with vemurafenib in melanoma Dai B, Kang SH, Gong W, Liu M, Aldape KD, Sawaya R, Huang with BRAF V600E mutation. N Engl J Med 364: 2507–2516. S. 2007. Aberrant FoxM1B expression increases matrix Chen R, Nishimura MC, Bumbaca SM, Kharbanda S, Forrest metalloproteinase-2 transcription and enhances the invasion WF, Kasman IM, Greve JM, Soriano RH, Gilmour LL, Rivers CS, et al. 2010. A hierarchy of self-renewing tumor-initiating of glioma cells. Oncogene 26: 6212–6219. Dai B, Pieper RO, Li D, Wei P, Liu M, Woo SY, Aldape KD, cell types in glioblastoma. Cancer Cell 17: 362–375. Chetty C, Vanamala SK, Gondi CS, Dinh DH, Gujrati M, Rao JS. Sawaya R, Xie K, Huang S. 2010. FoxM1B regulates NEDD4- 2012. MMP-9 induces CD44 cleavage and CD44 mediated 1 expression, leading to cellular transformation and full cell migration in glioblastoma xenograft cells. Cell Signal 24: malignant phenotype in immortalized human astrocytes. 549–559. Cancer Res 70: 2951–2961. Cheung HW, Cowley GS, Weir BA, Boehm JS, Rusin S, Scott JA, Dandy W. 1928. Removal of right cerebral hemisphere for East A, Ali LD, Lizotte PH, Wong TC, et al. 2011. Systematic certain tumors with hemiplegia. JAMA 90: 823–825. investigation of genetic vulnerabilities across cancer cell Dang L, White DW, Gross S, Bennett BD, Bittinger MA, Driggers lines reveals lineage-specific dependencies in . EM, Fantin VR, Jang HG, Jin S, Keenan MC, et al. 2010. Proc Natl Acad Sci 108: 12372–12377. Cancer-associated IDH1 mutations produce 2-hydroxygluta- Chi AS, Batchelor TT, Kwak EL, Clark JW, Wang DL, Wilner rate. Nature 465: 966. KD, Louis DN, Iafrate AJ. 2012. Rapid radiographic and Davis FG, Kupelian V, Freels S, McCarthy B, Surawicz T. 2001. clinical improvement after treatment of a MET-amplified Prevalence estimates for primary brain tumors in the United recurrent glioblastoma with a mesenchymal–epithelial tran- States by behavior and major histology groups. Neuro Oncol sition inhibitor. J Clin Oncol 30: e30–e33. doi: 10.1200/ 3: 152–158. JCO.2011.38.4586. Dawson MR, Polito A, Levine JM, Reynolds R. 2003. NG2- Chin L, Hahn WC, Getz G, Meyerson M. 2011. Making sense of expressing glial progenitor cells: An abundant and wide- cancer genomic data. Genes Dev 25: 534–555. spread population of cycling cells in the adult rat CNS. Chow LM, Endersby R, Zhu X, Rankin S, Qu C, Zhang J, Mol Cell Neurosci 24: 476–488. Broniscer A, Ellison DW, Baker SJ. 2011. Cooperativity De Raedt T, Walton Z, Yecies JL, Li D, Chen Y, Malone CF, within and among Pten, p53, and Rb pathways induces Maertens O, Jeong SM, Bronson RT, Lebleu V, et al. 2011. high-grade astrocytoma in adult brain. Cancer Cell 19: Exploiting cancer cell vulnerabilities to develop a combina- 305–316. tion therapy for ras-driven tumors. Cancer Cell 20: 400–413. Chowdhury R, Yeoh KK, Tian YM, Hillringhaus L, Bagg EA, Desai S, Pillai P, Win-Piazza H, Acevedo-Duncan M. 2011. PKC-i Rose NR, Leung IK, Li XS, Woon EC, Yang M, et al. 2011. promotes glioblastoma cell survival by phosphorylating and The oncometabolite 2-hydroxyglutarate inhibits histone inhibiting BAD through a phosphatidylinositol 3-kinase path- lysine demethylases. EMBO Rep 12: 463–469. way. Biochim Biophys Acta 1813: 1190–1197.

GENES & DEVELOPMENT 775 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al.

Desgrosellier JS, Cheresh DA. 2010. Integrins in cancer: Bi- Fan QW, Cheng C, Hackett C, Feldman M, Houseman BT, ological implications and therapeutic opportunities. Nat Rev Nicolaides T, Haas-Kogan D, James CD, Oakes SA, Debnath Cancer 10: 9–22. J, et al. 2010. Akt and autophagy cooperate to promote De Witt Hamer PC. 2010. Small molecule kinase inhibitors in survival of drug-resistant glioma. Sci Signal 3: ra81. doi: glioblastoma: A systematic review of clinical studies. Neuro 10.1126/scisignal.2001017. Oncol 12: 304–316. Fang M, Shen Z, Huang S, Zhao L, Chen S, Mak TW, Wang X. Dias-Santagata D, Akhavanfard S, David SS, Vernovsky K, 2010. The ER UDPase ENTPD5 promotes protein N-glyco- Kuhlmann G, Boisvert SL, Stubbs H, McDermott U, Settle- sylation, the Warburg effect, and proliferation in the PTEN man J, Kwak EL et al. 2010. Rapid targeted mutational pathway. Cell 143: 711–724. analysis of human tumours: A clinical platform to guide Figueroa ME, Abdel-Wahab O, Lu C, Ward PS, Patel J, Shih A, Li personalized cancer medicine. EMBO Mol Med 2: 146–158. Y, Bhagwat N, Vasanthakumar A, Fernandez HF, et al. 2010. Dias-Santagata D, Lam Q, Vernovsky K, Vena N, Lennerz JK, Leukemic IDH1 and IDH2 mutations result in a hypermeth- Borger DR, Batchelor TT, Ligon KL, Iafrate AJ, Ligon AH, ylation phenotype, disrupt TET2 function, and impair he- et al. 2011. BRAF V600E mutations are common in pleomor- matopoietic differentiation. Cancer Cell 18: 553–567. phic xanthoastrocytoma: Diagnostic and therapeutic implica- Fleming TP, Saxena A, Clark WC, Robertson JT, Oldfield EH, tions. PLoS ONE 6: e17948. doi: 10.1371/journal.pone.0017948. Aaronson SA, Ali IU. 1992. Amplification and/or overexpres- Dirks PB. 2010. Brain tumor stem cells: The cancer stem cell sion of platelet-derived growth factor receptors and epider- hypothesis writ large. Mol Oncol 4: 420–430. mal growth factor receptor in human glial tumors. Cancer Di Rocco F, Carroll RS, Zhang J, Black PM. 1998. Platelet- Res 52: 4550–4553. derived growth factor and its receptor expression in human Folkman J. 1971. Tumor angiogenesis: Therapeutic implica- oligodendrogliomas. Neurosurgery 42: 341–346. tions. N Engl J Med 285: 1182–1186. Druker BJ, Talpaz M, Resta DJ, Peng B, Buchdunger E, Ford JM, Fomchenko EI, Dougherty JD, Helmy KY, Katz AM, Pietras A, Lydon NB, Kantarjian H, Capdeville R, Ohno-Jones S, et al. Brennan C, Huse JT, Milosevic A, Holland EC. 2011. 2001. Efficacy and safety of a specific inhibitor of the BCR- Recruited cells can become transformed and overtake ABL tyrosine kinase in chronic myeloid leukemia. N Engl J PDGF-induced murine gliomas in vivo during tumor pro- Med 344: 1031–1037. gression. PLoS ONE 6: e20605. doi: 10.1371/journal.pone. Du R, Lu KV, Petritsch C, Liu P, Ganss R, Passegue E, Song H, 0020605. Vandenberg S, Johnson RS, Werb Z, et al. 2008. HIF1a induces Forsyth PA, Wong H, Laing TD, Rewcastle NB, Morris DG, the recruitment of bone marrow-derived vascular modulatory Muzik H, Leco KJ, Johnston RN, Brasher PM, Sutherland G, cells to regulate tumor angiogenesis and invasion. Cancer et al. 1999. Gelatinase-A (MMP-2), gelatinase-B (MMP-9) and Cell 13: 206–220. membrane type matrix metalloproteinase-1 (MT1-MMP) are Du J, Bernasconi P, Clauser KR, Mani DR, Finn SP, Beroukhim involved in different aspects of the pathophysiology of R, Burns M, Julian B, Peng XP, Hieronymus H, et al. 2009. malignant gliomas. Br J Cancer 79: 1828–1835. Bead-based profiling of tyrosine kinase phosphorylation iden- Freije WA, Castro-Vargas FE, Fang Z, Horvath S, Cloughesy T, tifies SRC as a potential target for glioblastoma therapy. Nat Liau LM, Mischel PS, Nelson SF. 2004. Gene expression Biotechnol 27: 77–83. profiling of gliomas strongly predicts survival. Cancer Res Dunn GP, Dunn IF, Curry WT. 2007. Focus on TILs: Prognostic 64: 6503–6510. significance of tumor infiltrating lymphocytes in human Friedman HS, Prados MD, Wen PY, Mikkelsen T, Schiff D, Abrey glioma. Cancer Immun 7: 12–28. LE, Yung WK, Paleologos N, Nicholas MK, Jensen R, et al. Dutil EM, Toker A, Newton AC. 1998. Regulation of conven- 2009. Bevacizumab alone and in combination with irinotecan tional protein kinase C isozymes by phosphoinositide-de- in recurrent glioblastoma. JClinOncol27: 4733–4740. pendent kinase 1 (PDK-1). Curr Biol 8: 1366–1375. Gallia GL, Rand V, Siu I-M, Eberhart CG, James CD, Marie SKN, Ekstrand AJ, James CD, Cavenee WK, Seliger B, Pettersson RF, Oba-Shinjo SM, Carlotti CG, Caballero OL, Simpson AJG, Collins VP. 1991. Genes for epidermal growth factor re- et al. 2006. PIK3CA gene mutations in pediatric and adult ceptor, transforming growth factor a, and epidermal growth glioblastoma multiforme. Mol Cancer Res 4: 709–714. factor and their expression in human gliomas in vivo. Cancer Gallia GL, Tyler BM, Hann CL, Siu IM, Giranda VL, Vescovi AL, Res 51: 2164–2172. Brem H, Riggins GJ. 2009. Inhibition of Akt inhibits growth Ellis LM, Hicklin DJ. 2008. Pathways mediating resistance to of glioblastoma and glioblastoma stem-like cells. Mol Cancer vascular endothelial growth factor-targeted therapy. Clin Ther 8: 386–393. Cancer Res 14: 6371–6375. Geha S, Pallud J, Junier MP, Devaux B, Leonard N, Chassoux F, Engelman JA, Luo J, Cantley LC. 2006. The evolution of Chneiweiss H, Daumas-Duport C, Varlet P. 2010. NG2+/ phosphatidylinositol 3-kinases as regulators of growth and Olig2+ cells are the major cycle-related cell population of the metabolism. Nat Rev Genet 7: 606–619. adult human normal brain. Brain Pathol 20: 399–411. Engelman JA, Zejnullahu K, Mitsudomi T, Song Y, Hyland C, Gerstner ER, Duda DG, di Tomaso E, Ryg PA, Loeffler JS, Park JO, Lindeman N, Gale CM, Zhao X, Christensen J, et al. Sorensen AG, Ivy P, Jain RK, Batchelor TT. 2009. VEGF 2007. MET amplification leads to gefitinib resistance in lung inhibitors in the treatment of cerebral edema in patients cancer by activating ERBB3 signaling. Science 316: 1039– with brain cancer. Nat Rev Clin Oncol 6: 229–236. 1043. Giese A, Westphal M. 1996. Glioma invasion in the central Fan QW, Knight ZA, Goldenberg DD, Yu W, Mostov KE, Stokoe nervous system. Neurosurgery 39: 235–252. D, Shokat KM, Weiss WA. 2006. A dual PI3 kinase/mTOR Gilbertson RJ, Rich JN. 2007. Making a tumour’s bed: Glioblas- inhibitor reveals emergent efficacy in glioma. Cancer Cell 9: toma stem cells and the vascular niche. Nat Rev Cancer 7: 341–349. 733–736. Fan QW, Cheng C, Knight ZA, Haas-Kogan D, Stokoe D, James Godar S, Ince TA, Bell GW, Feldser D, Donaher JL, Bergh J, Liu CD, McCormick F, Shokat KM, Weiss WA. 2009. EGFR A, Miu K, Watnick RS, Reinhardt F, et al. 2008. Growth- signals to mTOR through PKC and independently of Akt in inhibitory and tumor-suppressive functions of p53 depend on glioma. Sci Signal 2: ra4. doi: 10.1126/scisignal.2000014. its repression of CD44 expression. Cell 134: 62–73.

776 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma

Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, factor and its receptors in human glioma tissue: Expression Mesirov JP, Coller H, Loh ML, Downing JR, Caligiuri MA, of messenger RNA and protein suggests the presence of et al. 1999. Molecular classification of cancer: Class discov- autocrine and paracrine loops. Cancer Res 52: 3213–3219. ery and class prediction by gene expression monitoring. Hesselager G, Holland EC. 2003. Using mice to decipher the Science 286: 531–537. molecular genetics of brain tumors. Neurosurgery 53: 685– Graubert TA, Mardis ER. 2011. Genomics of acute myeloid 695. leukemia. Cancer J 17: 487–491. Hicklin DJ, Ellis LM. 2005. Role of the vascular endothelial Greenman C, Stephens P, Smith R, Dalgliesh GL, Hunter C, growth factor pathway in tumor growth and angiogenesis. Bignell G, Davies H, Teague J, Butler A, Stevens C, et al. J Clin Oncol 23: 1011–1027. 2007. Patterns of somatic mutation in human cancer ge- Hochberg FH, Pruitt A. 1980. Assumptions in the radiotherapy nomes. Nature 446: 153–158. of glioblastoma. Neurology 30: 907–911. Gritsenko PG, Ilina O, Friedl P. 2012. Interstitial guidance of Hoffman HJ, Duffner PK. 1985. Extraneural metastases of cancer invasion. J Pathol 226: 185–199. central nervous system tumors. Cancer 56: 1778–1782. Gross S, Cairns RA, Minden MD, Driggers EM, Bittinger MA, Holland EC, Hively WP, DePinho RA, Varmus HE. 1998. A Jang HG, Sasaki M, Jin S, Schenkein DP, Su SM, et al. 2010. constitutively active epidermal growth factor receptor co- Cancer-associated metabolite 2-hydroxyglutarate accumu- operates with disruption of G1 cell-cycle arrest pathways to lates in acute myelogenous leukemia with isocitrate de- induce glioma-like lesions in mice. Genes Dev 12: 3675– hydrogenase 1 and 2 mutations. J Exp Med 207: 339–344. 3685. Gulati N, Karsy M, Albert L, Murali R, Jhanwar-Uniyal M. 2009. Horbinski C, Hamilton RL, Nikiforov Y, Pollack IF. 2010. Involvement of mTORC1 and mTORC2 in regulation of Association of molecular alterations, including BRAF, with glioblastoma multiforme growth and motility. Int J Oncol biology and outcome in pilocytic astrocytomas. Acta Neuro- 35: 731–740. pathol 119: 641–649. Guo P, Hu B, Gu W, Xu L, Wang D, Huang HJ, Cavenee WK, Huang HS, Nagane M, Klingbeil CK, Lin H, Nishikawa R, Ji XD, Cheng SY. 2003. Platelet-derived growth factor-B enhances Huang CM, Gill GN, Wiley HS, Cavenee WK. 1997. The glioma angiogenesis by stimulating vascular endothelial enhanced tumorigenic activity of a mutant epidermal growth growth factor expression in tumor endothelia and by pro- factor receptor common in human cancers is mediated by moting pericyte recruitment. Am J Pathol 162: 1083–1093. threshold levels of constitutive tyrosine phosphorylation and Guryanova OA, Wu Q, Cheng L, Lathia JD, Huang Z, Yang J, unattenuated signaling. JBiolChem272: 2927–2935. MacSwords J, Eyler CE, McLendon RE, Heddleston JM, et al. Huang PH, Mukasa A, Bonavia R, Flynn RA, Brewer ZE, 2011. Nonreceptor tyrosine kinase BMX maintains self- Cavenee WK, Furnari FB, White FM. 2007. Quantitative renewal and tumorigenic potential of glioblastoma stem analysis of EGFRvIII cellular signaling networks reveals cells by activating STAT3. Cancer Cell 19: 498–511. a combinatorial therapeutic strategy for glioblastoma. Proc Hartmann C, Meyer J, Balss J, Capper D, Mueller W, Christians A, Natl Acad Sci 104: 12867–12872. Felsberg J, Wolter M, Mawrin C, Wick W, et al. 2009. Type and Huber JD, Egleton RD, Davis TP. 2001. Molecular physiology frequency of IDH1 and IDH2 mutations are related to astro- and pathophysiology of tight junctions in the blood–brain cytic and oligodendroglial differentiation and age: A study of barrier. Trends Neurosci 24: 719–725. 1,010 diffuse gliomas. Acta Neuropathol 118: 469–474. Hurtt MR, Moossy J, Donovan-Peluso M, Locker J. 1992. Hartmann C, Hentschel B, Wick W, Capper D, Felsberg J, Simon Amplification of epidermal growth factor receptor gene in M, Westphal M, Schackert G, Meyermann R, Pietsch T, et al. gliomas: Histopathology and prognosis. J Neuropathol Exp 2010. Patients with IDH1 wild type anaplastic astrocytomas Neurol 51: 84–90. exhibit worse prognosis than IDH1-mutated glioblastomas, Huse JT, Phillips HS, Brennan CW. 2011. Molecular subclassi- and IDH1 mutation status accounts for the unfavorable fication of diffuse gliomas: Seeing order in the chaos. Glia 59: prognostic effect of higher age: Implications for classification 1190–1199. of gliomas. Acta Neuropathol 120: 707–718. Ichimura K, Pearson DM, Kocialkowski S, Backlund LM, Chan Hawkins C, Walker E, Mohamed N, Zhang C, Jacob K, Shirinian R, Jones DT, Collins VP. 2009. IDH1 mutations are present in M, Alon N, Kahn D, Fried I, Scheinemann K, et al. 2011. BRAF– the majority of common adult gliomas but rare in primary KIAA1549 fusion predicts better clinical outcome in pediatric glioblastomas. Neuro Oncol 11: 341–347. low-grade astrocytoma. Clin Cancer Res 17: 4790–4798. Idbaih A, Carvalho Silva R, Criniere E, Marie Y, Carpentier C, Hegi ME, Diserens AC, Gorlia T, Hamou MF, de Tribolet N, Boisselier B, Taillibert S, Rousseau A, Mokhtari K, Ducray F, Weller M, Kros JM, Hainfellner JA, Mason W, Mariani L, et al. 2008. Genomic changes in progression of low-grade et al. 2005. MGMT gene silencing and benefit from temozo- gliomas. J Neurooncol 90: 133–140. lomide in glioblastoma. N Engl J Med 352: 997–1003. Ignatova TN, Kukekov VG, Laywell ED, Suslov ON, Vrionis FD, Heimberger AB, Hlatky R, Suki D, Yang D, Weinberg J, Gilbert Steindler DA. 2002. Human cortical glial tumors contain M, Sawaya R, Aldape K. 2005. Prognostic effect of epidermal neural stem-like cells expressing astroglial and neuronal growth factor receptor and EGFRvIII in glioblastoma multi- markers in vitro. Glia 39: 193–206. forme patients. Clin Cancer Res 11: 1462–1466. Inda MM, Bonavia R, Mukasa A, Narita Y, Sah DW, Vandenberg Hemmati HD, Nakano I, Lazareff JA, Masterman-Smith M, S, Brennan C, Johns TG, Bachoo R, Hadwiger P, et al. 2010. Geschwind DH, Bronner-Fraser M, Kornblum HI. 2003. Tumor heterogeneity is an active process maintained by Cancerous stem cells can arise from pediatric brain tumors. a mutant EGFR-induced cytokine circuit in glioblastoma. Proc Natl Acad Sci 100: 15178–15183. Genes Dev 24: 1731–1745. Henson JW, Schnitker BL, Correa KM, von Deimling A, Fassbender Iwamoto FM, Abrey LE, Beal K, Gutin PH, Rosenblum MK, F, Xu HJ, Benedict WF, Yandell DW, Louis DN. 1994. The Reuter VE, DeAngelis LM, Lassman AB. 2009. Patterns of retinoblastoma gene is involved in malignant progression relapse and prognosis after bevacizumab failure in recurrent of astrocytomas. Ann Neurol 36: 714–721. glioblastoma. Neurology 73: 1200–1206. Hermanson M, Funa K, Hartman M, Claesson-Welsh L, Heldin Iwamoto FM, Hottinger AF, Karimi S, Riedel E, Dantis J, Jahdi CH, Westermark B, Nister M. 1992. Platelet-derived growth M, Panageas KS, Lassman AB, Abrey LE, Fleisher M, et al.

GENES & DEVELOPMENT 777 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al.

2011. Serum YKL-40 is a marker of prognosis and disease AM, de Klerk JB, et al. 2010. IDH2 mutations in patients status in high-grade gliomas. Neuro Oncol 13: 1244–1251. with D-2-hydroxyglutaric aciduria. Science 330: 336. Jain RK. 2005. Normalization of tumor vasculature: An emerg- Kreisl TN, Kim L, Moore K, Duic P, Royce C, Stroud I, Garren ing concept in antiangiogenic therapy. Science 307: 58–62. N, Mackey M, Butman JA, Camphausen K, et al. 2009. Phase Janne PA, Engelman JA, Johnson BE. 2005. Epidermal growth II trial of single-agent bevacizumab followed by bevacizumab factor receptor mutations in non-small-cell lung cancer: plus irinotecan at tumor progression in recurrent glioblas- Implications for treatment and tumor biology. J Clin Oncol toma. J Clin Oncol 27: 740–745. 23: 3227–3234. Kwak EL, Bang YJ, Camidge DR, Shaw AT, Solomon B, Maki Jaros E, Perry RH, Adam L, Kelly PJ, Crawford PJ, Kalbag RM, RG, Ou SH, Dezube BJ, Janne PA, Costa DB, et al. 2010. Mendelow AD, Sengupta RP, Pearson AD. 1992. Prognostic Anaplastic lymphoma kinase inhibition in non-small-cell implications of p53 protein, epidermal growth factor re- lung cancer. N Engl J Med 363: 1693–1703. ceptor, and Ki-67 labelling in brain tumours. Br J Cancer 66: Kwok JC, Dick G, Wang D, Fawcett JW. 2011. Extracellular 373–385. matrix and perineuronal nets in CNS repair. Dev Neurobiol Jo M, Stolz DB, Esplen JE, Dorko K, Michalopoulos GK, Strom 71: 1073–1089. SC. 2000. Cross-talk between epidermal growth factor re- Kwon CH, Zhao D, Chen J, Alcantara S, Li Y, Burns DK, Mason ceptor and c-Met signal pathways in transformed cells. J Biol RP, Lee EY, Wu H, Parada LF. 2008. Pten haploinsufficiency Chem 275: 8806–8811. accelerates formation of high-grade astrocytomas. Cancer Jones DT, Kocialkowski S, Liu L, Pearson DM, Backlund LM, Res 68: 3286–3294. Ichimura K, Collins VP. 2008. Tandem duplication producing Lai A, Kharbanda S, Pope WB, Tran A, Solis OE, Peale F, Forrest a novel oncogenic BRAF fusion gene defines the majority of WF, Pujara K, Carrillo JA, Pandita A, et al. 2011. Evidence for pilocytic astrocytomas. Cancer Res 68: 8673–8677. sequenced molecular evolution of IDH1 mutant glioblastoma Jones DT, Kocialkowski S, Liu L, Pearson DM, Ichimura K, from a distinct cell of origin. JClinOncol29: 4482–4490. Collins VP. 2009. Oncogenic RAF1 rearrangement and a novel Lal B, Goodwin CR, Sang Y, Foss CA, Cornet K, Muzamil S, BRAF mutation as alternatives to KIAA1549:BRAF fusion in Pomper MG, Kim J, Laterra J. 2009. EGFRvIII and c-Met activating the MAPK pathway in pilocytic astrocytoma. pathway inhibitors synergize against PTEN-null/EGFRvIII+ Oncogene 28: 2119–2123. glioblastoma xenografts. Mol Cancer Ther 8: 1751–1760. Kaelin WG Jr. 2011. Cancer and altered metabolism: Potential Lathia JD, Heddleston JM, Venere M, Rich JN. 2011a. Deadly importance of hypoxia-inducible factor and 2-oxoglutarate- teamwork: Neural cancer stem cells and the tumor micro- dependent dioxygenases. Cold Spring Harb Symp Quant Biol environment. Cell Stem Cell 8: 482–485. doi: 10.1101/sqb.2011.76.010975. Lathia JD, Venere M, Rao MS, Rich JN. 2011b. Seeing is Kargiotis O, Rao JS, Kyritsis AP. 2006. Mechanisms of angio- believing: Are cancer stem cells the Loch Ness monster of genesis in gliomas. J Neurooncol 78: 281–293. tumor biology? Stem Cell Rev 7: 227–237. Kaur B, Khwaja FW, Severson EA, Matheny SL, Brat DJ, Van Lee J, Kotliarova S, Kotliarov Y, Li A, Su Q, Donin NM, Meir EG. 2005. Hypoxia and the hypoxia-inducible-factor Pastorino S, Purow BW, Christopher N, Zhang W, et al. pathway in glioma growth and angiogenesis. Neuro Oncol 7: 2006a. Tumor stem cells derived from glioblastomas cul- 134–153. tured in bFGF and EGF more closely mirror the phenotype Kerbel RS. 2008. Tumor angiogenesis. N Engl J Med 358: 2039– and genotype of primary tumors than do serum-cultured cell 2049. lines. Cancer Cell 9: 391–403. Kita D, Yonekawa Y, Weller M, Ohgaki H. 2007. PIK3CA Lee JC, Vivanco I, Beroukhim R, Huang JH, Feng WL, DeBiasi RM, alterations in primary (de novo) and secondary glioblasto- Yoshimoto K, King JC, Nghiemphu P, Yuza Y, et al. 2006b. mas. Acta Neuropathol 113: 295–302. Epidermal growth factor receptor activation in glioblastoma Kleber S, Sancho-Martinez I, Wiestler B, Beisel A, Gieffers C, through novel missense mutations in the extracellular domain. Hill O, Thiemann M, Mueller W, Sykora J, Kuhn A, et al. PLoS Med 3: e485. doi: 10.1371/journal.pmed.0030485. 2008. Yes and PI3K bind CD95 to signal invasion of glioblas- Le Good JA, Ziegler WH, Parekh DB, Alessi DR, Cohen P, Parker toma. Cancer Cell 13: 235–248. PJ. 1998. Protein kinase C isotypes controlled by phospho- Koga K, Todaka T, Morioka M, Hamada J, Kai Y, Yano S, inositide 3-kinase through the protein kinase PDK1. Science Okamura A, Takakura N, Suda T, Ushio Y. 2001. Expression 281: 2042–2045. of angiopoietin-2 in human glioma cells and its role for Levin VA, Phuphanich S, Yung WK, Forsyth PA, Maestro RD, angiogenesis. Cancer Res 61: 6248–6254. Perry JR, Fuller GN, Baillet M. 2006. Randomized, double- Koivunen P, Lee S, Duncan CG, Lopez G, Lu G, Ramkissoon S, blind, placebo-controlled trial of marimastat in glioblastoma Losman JA, Joensuu P, Bergmann U, Gross S, et al. 2012. multiforme patients following surgery and irradiation. Transformation by the (R)-enantiomer of 2-hydroxyglutarate J Neurooncol 78: 295–302. linked to EGLN activation. Nature doi: 10.1038/nature10898. Levitt RJ, Georgescu MM, Pollak M. 2005. PTEN-induction in Kong DS, Song SY, Kim DH, Joo KM, Yoo JS, Koh JS, Dong U251 glioma cells decreases the expression of insulin-like SM, Suh YL, Lee JI, Park K, et al. 2009. Prognostic sig- growth factor binding protein-2. Biochem Biophys Res Com- nificance of c-Met expression in glioblastomas. Cancer mun 336: 1056–1061. 115: 140–148. Li M, Mukasa A, Del-Mar Inda M, Zhang J, Chin L, Cavenee W, Kotliarov Y, Steed ME, Christopher N, Walling J, Su Q, Center Furnari F. 2011. Guanylate binding protein 1 is a novel A, Heiss J, Rosenblum M, Mikkelsen T, Zenklusen JC, et al. effector of EGFR-driven invasion in glioblastoma. J Exp 2006. High-resolution global genomic survey of 178 gliomas Med 208: 2657–2673. reveals novel regions of copy number alteration and allelic Ligon KL, Alberta JA, Kho AT, Weiss J, Kwaan MR, Nutt CL, imbalances. Cancer Res 66: 9428–9436. Louis DN, Stiles CD, Rowitch DH. 2004. The oligodendrog- Koul D. 2008. PTEN signaling pathways in glioblastoma. lial lineage marker OLIG2 is universally expressed in diffuse Cancer Biol Ther 7: 1321–1325. gliomas. J Neuropathol Exp Neurol 63: 499–509. Kranendijk M, Struys EA, van Schaftingen E, Gibson KM, Ligon KL, Huillard E, Mehta S, Kesari S, Liu H, Alberta JA, Kanhai WA, van der Knaap MS, Amiel J, Buist NR, Das Bachoo RM, Kane M, Louis DN, Depinho RA, et al. 2007.

778 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma

Olig2-regulated lineage-restricted pathway controls replica- Maccario H, Perera NM, Davidson L, Downes CP, Leslie NR. tion competence in neural stem cells and malignant glioma. 2007. PTEN is destabilized by phosphorylation on Thr366. Neuron 53: 503–517. Biochem J 405: 439–444. Lim SK, Llaguno SR, McKay RM, Parada LF. 2011. Glioblastoma MacConaill LE, Campbell CD, Kehoe SM, Bass AJ, Hatton C, multiforme: A perspective on recent findings in human Niu L, Davis M, Yao K, Hanna M, Mondal C, et al. 2009. cancer and mouse models. BMB reports 44: 158–164. Profiling critical cancer gene mutations in clinical tumor Lino MM, Merlo A. 2011. PI3Kinase signaling in glioblastoma. samples. PLoS ONE 4: e7887. doi: 10.1371/journal.pone. J Neurooncol 103: 417–427. 0007887. Lipinski CA, Tran NL, Viso C, Kloss J, Yang Z, Berens ME, Maher EA, Brennan C, Wen PY, Durso L, Ligon KL, Richardson Loftus JC. 2008. Extended survival of Pyk2 or FAK deficient A, Khatry D, Feng B, Sinha R, Louis DN, et al. 2006. Marked orthotopic glioma xenografts. J Neurooncol 90: 181–189. genomic differences characterize primary and secondary Liu C, Sage JC, Miller MR, Verhaak RG, Hippenmeyer S, Vogel glioblastoma subtypes and identify two distinct molecular H, Foreman O, Bronson RT, Nishiyama A, Luo L, et al. 2011. and clinical secondary glioblastoma entities. Cancer Res 66: Mosaic analysis with double markers reveals tumor cell of 11502–11513. origin in glioma. Cell 146: 209–221. Maity A, Pore N, Lee J, Solomon D, O’Rourke DM. 2000. Liu Q, Wang J, Kang SA, Thoreen CC, Hur W, Ahmed T, Sabatini Epidermal growth factor receptor transcriptionally up-regulates DM, Gray NS. 2011. Discovery of 9-(6-aminopyridin-3-yl)-1- vascular endothelial growth factor expression in human glio- (3-(trifluoromethyl)phenyl)benzo[h][1,6]naphthyr idin-2(1H)- blastoma cells via a pathway involving phosphatidylinositol one (Torin2) as a potent, selective, and orally available mam- 39-kinase and distinct from that induced by hypoxia. Cancer malian target of rapamycin (mTOR) inhibitor for treatment of Res 60: 5879–5886. cancer. J Med Chem 54: 1473–1480. Mallory F. 1914. The principles of pathologic histology. Saun- Loenarz C, Schofield CJ. 2008. Expanding chemical biology of ders, Philadelphia. 2-oxoglutarate oxygenases. Nat Chem Biol 4: 152–156. Mardis ER, Ding L, Dooling DJ, Larson DE, McLellan MD, Chen Lokker NA, Sullivan CM, Hollenbach SJ, Israel MA, Giese NA. K, Koboldt DC, Fulton RS, Delehaunty KD, McGrath SD, 2002. Platelet-derived growth factor (PDGF) autocrine sig- et al. 2009. Recurring mutations found by sequencing an naling regulates survival and mitogenic pathways in glio- acute myeloid leukemia genome. N Engl J Med 361: 1058– blastoma cells: Evidence that the novel PDGF-C and PDGF- 1066. D ligands may play a role in the development of brain Matsukado Y, Maccarty CS, Kernohan JW. 1961. The growth of tumors. Cancer Res 62: 3729–3735. glioblastoma multiforme (astrocytomas, grades 3 and 4) in Long DM. 1970. Capillary ultrastructure and the blood–brain neurosurgical practice. J Neurosurg 18: 636–644. barrier in human malignant brain tumors. J Neurosurg 32: McGillicuddy LT, Fromm JA, Hollstein PE, Kubek S, Beroukhim 127–144. R, De Raedt T, Johnson BW, Williams SM, Nghiemphu P, Lopez KA, Tannenbaum AM, Assanah MC, Linskey K, Yun J, Liau LM, et al. 2009. Proteasomal and genetic inactivation of Kangarlu A, Gil OD, Canoll P, Bruce JN. 2011. Convection- the NF1 tumor suppressor in gliomagenesis. Cancer Cell 16: enhanced delivery of topotecan into a PDGF-driven model of 44–54. glioblastoma prolongs survival and ablates both tumor- Mellinghoff IK, Wang MY, Vivanco I, Haas-Kogan DA, Zhu S, Dia initiating cells and recruited glial progenitors. Cancer Res EQ, Lu KV, Yoshimoto K, Huang JH, Chute DJ, et al. 2005. 71: 3963–3971. Molecular determinants of the response of glioblastomas to Louis DN. 1994. The p53 gene and protein in human brain EGFR kinase inhibitors. N Engl J Med 353: 2012–2024. tumors. J Neuropathol Exp Neurol 53: 11–21. Mellinghoff IK, Cloughesy TF, Mischel PS. 2007. PTEN-mediated Louis DN. 2006. Molecular pathology of malignant gliomas. resistance to epidermal growth factor receptor kinase in- Annu Rev Pathol 1: 97–117. hibitors. Clin Cancer Res 13: 378–381. Louis DN, Ohgaki H, Wiestler OD, Cavenee WK, Burger PC, Mermel CH, Schumacher SE, Hill B, Meyerson ML, Beroukhim Jouvet A, Scheithauer BW, Kleihues P. 2007. The 2007 WHO R, Getz G. 2011. GISTIC2.0 facilitates sensitive and confi- classification of tumours of the central nervous system. Acta dent localization of the targets of focal somatic copy-number Neuropathol 114: 97–109. alteration in human cancers. Genome Biol 12: R41. doi: Lu KV, Zhu S, Cvrljevic A, Huang TT, Sarkaria S, Ahkavan D, 10.1186/gb-2011-12-4-r41. Dang J, Dinca EB, Plaisier SB, Oderberg I, et al. 2009. Fyn and Metallo CM, Gameiro PA, Bell EL, Mattaini KR, Yang J, Hiller SRC are effectors of oncogenic epidermal growth factor K, Jewell CM, Johnson ZR, Irvine DJ, Guarente L, et al. 2012. receptor signaling in glioblastoma patients. Cancer Res 69: Reductive glutamine metabolism by IDH1 mediates lipogen- 6889–6898. esis under hypoxia. Nature 481: 380–384. Lu C, Ward PS, Kapoor GS, Rohle D, Turcan S, Abdel-Wahab O, Meyerson M, Gabriel S, Getz G. 2010. Advances in understand- Edwards CR, Khanin R, Figueroa ME, Melnick A, et al. 2012. ing cancer genomes through second-generation sequencing. IDH mutation impairs histone demethylation and results Nat Rev Genet 11: 685–696. in a block to cell differentiation. Nature doi: 10.1038/ Miao H, Li DQ, Mukherjee A, Guo H, Petty A, Cutter J, Basilion nature10860. JP, Sedor J, Wu J, Danielpour D, et al. 2009. EphA2 mediates Luchman HA, Stechishin OD, Dang NH, Blough MD, Chesnelong ligand-dependent inhibition and ligand-independent promo- C, Kelly JJ, Nguyen SA, Chan JA, Weljie AM, Cairncross JG, tion of cell migration and invasion via a reciprocal regulatory et al. 2012. An in vivo patient-derived model of endogenous loop with Akt. Cancer Cell 16: 9–20. IDH1-mutant glioma. Neuro Oncol 14: 184–191. Miller CR, Perry A. 2007. Glioblastoma. Arch Pathol Lab Med Lynch TJ, Bell DW, Sordella R, Gurubhagavatula S, Okimoto 131: 397–406. RA, Brannigan BW, Harris PL, Haserlat SM, Supko JG, Mizoguchi M, Nutt CL, Mohapatra G, Louis DN. 2004. Genetic Haluska FG, et al. 2004. Activating mutations in the epider- alterations of phosphoinositide 3-kinase subunit genes in mal growth factor receptor underlying responsiveness of non- human glioblastomas. Brain Pathol 14: 372–377. small-cell lung cancer to gefitinib. N Engl J Med 350: 2129– Molina JR, Agarwal NK, Morales FC, Hayashi Y, Aldape KD, 2139. Cote G, Georgescu MM. 2012. PTEN, NHERF1 and PHLPP

GENES & DEVELOPMENT 779 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al.

form a tumor suppressor network that is disabled in glio- Ohgaki H, Kleihues P. 2005. Population-based studies on in- blastoma. Oncogene 31: 1264–1274. cidence, survival rates, and genetic alterations in astrocytic Murai T, Miyazaki Y, Nishinakamura H, Sugahara KN, Miyauchi and oligodendroglial gliomas. J Neuropathol Exp Neurol 64: T, Sako Y, Yanagida T, Miyasaka M. 2004. Engagement of 479–489. CD44 promotes Rac activation and CD44 cleavage during Ohgaki H, Kleihues P. 2009. Genetic alterations and signaling tumor cell migration. J Biol Chem 279: 4541–4550. pathways in the evolution of gliomas. Cancer Sci 100: 2235– Nabeshima K, Shimao Y, Sato S, Kataoka H, Moriyama T, 2241. Kawano H, Wakisaka S, Koono M. 1997. Expression of Onishi M, Ichikawa T, Kurozumi K, Date I. 2011. Angiogenesis c-Met correlates with grade of malignancy in human astro- and invasion in glioma. Brain Tumor Pathol 28: 13–24. cytic tumours: An immunohistochemical study. Histopa- Ozawa T, Brennan CW, Wang L, Squatrito M, Sasayama T, thology 31: 436–443. Nakada M, Huse JT, Pedraza A, Utsuki S, Yasui Y, et al. 2010. Nagane M, Levitzki A, Gazit A, Cavenee WK, Huang HJ. 1998. PDGFRA gene rearrangements are frequent genetic events in Drug resistance of human glioblastoma cells conferred by PDGFRA-amplified glioblastomas. Genes Dev 24: 2205– a tumor-specific mutant epidermal growth factor receptor 2218. through modulation of Bcl-XL and caspase-3-like proteases. Pandey A, Lazar DF, Saltiel AR, Dixit VM. 1994. Activation of the Proc Natl Acad Sci 95: 5724–5729. Eck receptor protein tyrosine kinase stimulates phosphatidy- Nakada M, Nakada S, Demuth T, Tran NL, Hoelzinger DB, linositol 3-kinase activity. JBiolChem269: 30154–30157. Berens ME. 2007. Molecular targets of glioma invasion. Cell Panner A, Crane CA, Weng C, Feletti A, Parsa AT, Pieper RO. Mol Life Sci 64: 458–478. 2009. A novel PTEN-dependent link to ubiquitination con- Narayana A, Kunnakkat SD, Medabalmi P, Golfinos J, Parker E, trols FLIPS stability and TRAIL sensitivity in glioblastoma Knopp E, Zagzag D, Eagan P, Gruber D, Gruber ML. 2012. multiforme. Cancer Res 69: 7911–7916. Change in pattern of relapse after antiangiogenic therapy in Pansuriya TC, van Eijk R, d’Adamo P, van Ruler MA, Kuijjer high-grade glioma. Int J Radiat Oncol Biol Phys 82: 77–82. ML, Oosting J, Cleton-Jansen AM, van Oosterwijk JG, Narita Y, Nagane M, Mishima K, Huang HJ, Furnari FB, Cavenee Verbeke SL, Meijer D, et al. 2011. Somatic mosaic IDH1 WK. 2002. Mutant epidermal growth factor receptor signaling and IDH2 mutations are associated with enchondroma and down-regulates p27 through activation of the phosphatidyli- spindle cell hemangioma in Ollier disease and Maffucci nositol 3-kinase/Akt pathway in glioblastomas. Cancer Res syndrome. Nat Genet 43: 1256–1261. 62: 6764–6769. Parsa AT, Wachhorst S, Lamborn KR, Prados MD, McDermott Natsume A, Kato T, Kinjo S, Enomoto A, Toda H, Shimato S, MW, Berger MS, Chang SM. 2005. Prognostic significance of Ohka F, Motomura K, Kondo Y, Miyata T, et al. 2011. Girdin intracranial dissemination of glioblastoma multiforme in maintains the stemness of glioblastoma stem cells. Onco- adults. J Neurosurg 102: 622–628. gene doi: 10.1038/onc.2011.466. Parsons DW, Jones S, Zhang X, Lin JC, Leary RJ, Angenendt P, Navin N, Hicks J. 2011. Future medical applications of single- Mankoo P, Carter H, Siu IM, Gallia GL, et al. 2008. An cell sequencing in cancer. Genome Med 3: 31. doi: 10.1186/ integrated genomic analysis of human glioblastoma multi- gm247. forme. Science 321: 1807–1812. Navin N, Kendall J, Troge J, Andrews P, Rodgers L, McIndoo J, Paschka P, Schlenk RF, Gaidzik VI, Habdank M, Kronke J, Cook K, Stepansky A, Levy D, Esposito D, et al. 2011. Bullinger L, Spath D, Kayser S, Zucknick M, Gotze K, Tumour evolution inferred by single-cell sequencing. Nature et al. 2010. IDH1 and IDH2 mutations are frequent genetic 472: 90–94. alterations in acute myeloid leukemia and confer adverse Nishikawa R, Sugiyama T, Narita Y, Furnari F, Cavenee WK, prognosis in cytogenetically normal acute myeloid leukemia Matsutani M. 2004. Immunohistochemical analysis of the with NPM1 mutation without FLT3 internal tandem dupli- mutant epidermal growth factor, DEGFR, in glioblastoma. cation. J Clin Oncol 28: 3636–3643. Brain Tumor Pathol 21: 53–56. Patenaude A, Parker J, Karsan A. 2010. Involvement of endo- Norden AD, Drappatz J, Wen PY. 2008a. Novel anti-angiogenic thelial progenitor cells in tumor vascularization. Microvasc therapies for malignant gliomas. Lancet Neurol 7: 1152– Res 79: 217–223. 1160. Paternot S, Roger PP. 2009. Combined inhibition of MEK and NordenAD,YoungGS,SetayeshK,MuzikanskyA,KlufasR, mammalian target of rapamycin abolishes phosphorylation Ross GL, Ciampa AS, Ebbeling LG, Levy B, Drappatz J, et al. of -dependent kinase 4 in glioblastoma cell lines and 2008b. Bevacizumab for recurrent malignant gliomas: Efficacy, prevents their proliferation. Cancer Res 69: 4577–4581. toxicity, and patterns of recurrence. Neurology 70: 779–787. Perren A, Komminoth P, Saremaslani P, Matter C, Feurer S, Lees Noushmehr H, Weisenberger DJ, Diefes K, Phillips HS, Pujara K, JA, Heitz PU, Eng C. 2000. Mutation and expression analyses Berman BP, Pan F, Pelloski CE, Sulman EP, Bhat KP, et al. reveal differential subcellular compartmentalization of 2010. Identification of a CpG island methylator phenotype PTEN in endocrine pancreatic tumors compared to normal that defines a distinct subgroup of glioma. Cancer Cell 17: islet cells. Am J Pathol 157: 1097–1103. 510–522. Persson AI, Petritsch C, Swartling FJ, Itsara M, Sim FJ, Auvergne Nutt CL, Mani DR, Betensky RA, Tamayo P, Cairncross JG, R, Goldenberg DD, Vandenberg SR, Nguyen KN, Yakovenko Ladd C, Pohl U, Hartmann C, McLaughlin ME, Batchelor TT, S, et al. 2010. Non-stem cell origin for oligodendroglioma. et al. 2003. Gene expression-based classification of malig- Cancer Cell 18: 669–682. nant gliomas correlates better with survival than histological Pfister S, Janzarik WG, Remke M, Ernst A, Werft W, Becker N, classification. Cancer Res 63: 1602–1607. Toedt G, Wittmann A, Kratz C, Olbrich H, et al. 2008. BRAF Nyberg P, Xie L, Kalluri R. 2005. Endogenous inhibitors of gene duplication constitutes a mechanism of MAPK path- angiogenesis. Cancer Res 65: 3967–3979. way activation in low-grade astrocytomas. J Clin Invest 118: Ogden AT, Waziri AE, Lochhead RA, Fusco D, Lopez K, Ellis JA, 1739–1749. Kang J, Assanah M, McKhann GM, Sisti MB et al. 2008. Phillips HS, Kharbanda S, Chen R, Forrest WF, Soriano RH, Wu Identification of A2B5+CD133À tumor-initiating cells in TD, Misra A, Nigro JM, Colman H, Soroceanu L, et al. 2006. adult human gliomas. Neurosurgery 62: 505–515. Molecular subclasses of high-grade glioma predict prognosis,

780 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma

delineate a pattern of disease progression, and resemble effects of isocitrate dehydrogenase 1 and 2 mutations on the stages in neurogenesis. Cancer Cell 9: 157–173. cellular metabolome. Proc Natl Acad Sci 108: 3270–3275. Phung TL, Ziv K, Dabydeen D, Eyiah-Mensah G, Riveros M, Reyes-Mugica M, Rieger-Christ K, Ohgaki H, Ekstrand BC, Perruzzi C, Sun J, Monahan-Earley RA, Shiojima I, Nagy JA, Helie M, Kleinman G, Yahanda A, Fearon ER, Kleihues P, et al. 2006. Pathological angiogenesis is induced by sustained Reale MA. 1997. Loss of DCC expression and glioma pro- Akt signaling and inhibited by rapamycin. Cancer Cell 10: gression. Cancer Res 57: 382–386. 159–170. Reznik TE, Sang Y, Ma Y, Abounader R, Rosen EM, Xia S, Pietrak B, Zhao H, Qi H, Quinn C, Gao E, Boyer JG, Concha N, Laterra J. 2008. Transcription-dependent epidermal growth Brown K, Duraiswami C, Wooster R, et al. 2011. A tale of factor receptor activation by hepatocyte growth factor. Mol two subunits: How the neomorphic R132H IDH1 mutation Cancer Res 6: 139–150. enhances production of aHG. Biochemistry 50: 4804–4812. Ricci-Vitiani L, Pallini R, Biffoni M, Todaro M, Invernici G, Pillay V, Allaf L, Wilding AL, Donoghue JF, Court NW, Greenall Cenci T, Maira G, Parati EA, Stassi G, Larocca LM, et al. SA, Scott AM, Johns TG. 2009. The plasticity of oncogene 2010. Tumour vascularization via endothelial differentiation addiction: Implications for targeted therapies directed to of glioblastoma stem-like cells. Nature 468: 824–828. receptor tyrosine kinases. Neoplasia 11: 448–458. Riddick G, Fine HA. 2011. Integration and analysis of genome- Pollard SM, Yoshikawa K, Clarke ID, Danovi D, Stricker S, scale data from gliomas. Nat Rev Neurol 7: 439–450. Russell R, Bayani J, Head R, Lee M, Bernstein M, et al. 2009. Riemenschneider MJ, Mueller W, Betensky RA, Mohapatra G, Glioma stem cell lines expanded in adherent culture have Louis DN. 2005. In situ analysis of integrin and growth factor tumor-specific phenotypes and are suitable for chemical and receptor signaling pathways in human glioblastomas sug- genetic screens. Cell Stem Cell 4: 568–580. gests overlapping relationships with focal adhesion kinase Pope WB, Prins RM, Albert Thomas M, Nagarajan R, Yen KE, activation. Am J Pathol 167: 1379–1387. Bittinger MA, Salamon N, Chou AP, Yong WH, Soto H, et al. Rodin SN, Rodin AS. 1998. Strand asymmetry of CpG transi- 2012. Non-invasive detection of 2-hydroxyglutarate and other tions as indicator of G1 phase-dependent origin of multiple metabolites in IDH1 mutant glioma patients using magnetic tumorigenic p53 mutations in stem cells. Proc Natl Acad Sci 11927–11932. resonance spectroscopy. JNeurooncol107: 197–205. 95: Pore N, Liu S, Haas-Kogan DA, O’Rourke DM, Maity A. 2003. Rodriguez-Viciana P, Warne PH, Dhand R, Vanhaesebroeck B, Gout I, Fry MJ, Waterfield MD, Downward J. 1994. Phos- PTEN mutation and epidermal growth factor receptor acti- phatidylinositol-3-OH kinase as a direct target of Ras. vation regulate vascular endothelial growth factor (VEGF) Nature 370: 527–532. mRNA expression in human glioblastoma cells by trans- Rodriguez-Viciana P, Warne PH, Vanhaesebroeck B, Waterfield activating the proximal VEGF promoter. Cancer Res 63: MD, Downward J. 1996. Activation of phosphoinositide 3- 236–241. kinase by interaction with Ras and by point mutation. Porter KR, McCarthy BJ, Freels S, Kim Y, Davis FG. 2010. EMBO J 15: 2442–2451. Prevalence estimates for primary brain tumors in the United Rothenberg SM, Engelman JA, Le S, Riese DJ II, Haber DA, States by age, gender, behavior, and histology. Neuro Oncol Settleman J. 2008. Modeling oncogene addiction using RNA 12: 520–527. interference. Proc Natl Acad Sci 105: 12480–12484. Quant EC, Norden AD, Drappatz J, Muzikansky A, Doherty L, Rutka JT, Apodaca G, Stern R, Rosenblum M. 1988. The Lafrankie D, Ciampa A, Kesari S, Wen PY. 2009. Role of extracellular matrix of the central and peripheral nervous a second chemotherapy in recurrent malignant glioma pa- systems: Structure and function. J Neurosurg 69: 155–170. tients who progress on bevacizumab. Neuro Oncol 11: 550– Rutka JT, Muller M, Hubbard SL, Forsdike J, Dirks PB, Jung S, 555. Tsugu A, Ivanchuk S, Costello P, Mondal S, et al. 1999. Raimundo N, Baysal BE, Shadel GS. 2011. Revisiting the TCA Astrocytoma adhesion to extracellular matrix: Functional cycle: Signaling to tumor formation. Trends Mol Med 17: significance of integrin and focal adhesion kinase expression. 641–649. J Neuropathol Exp Neurol 58: 198–209. Rao JS. 2003. Molecular mechanisms of glioma invasiveness: Sahm F, Capper D, Pusch S, Balss J, Koch A, Langhans CD, Okun The role of proteases. Nat Rev Cancer 3: 489–501. JG, von Deimling A. 2012. Detection of 2-hydroxyglutarate Rawlings CE III, Bullard DE, Burger PC, Friedman AH. 1987. A in formalin-fixed paraffin-embedded glioma specimens by case of Ollier’s disease associated with two intracranial gas chromatography/mass spectrometry. Brain Pathol 22: gliomas. Neurosurgery 21: 400–403. 26–31. Reardon DA, Neyns B, Weller M, Tonn JC, Nabors LB, Stupp R. Sanai N, Nguyen T, Ihrie RA, Mirzadeh Z, Tsai HH, Wong M, 2011a. Cilengitide: An RGD pentapeptide anb3 and anb5 Gupta N, Berger MS, Huang E, Garcia-Verdugo JM, et al. integrin inhibitor in development for glioblastoma and other 2011. Corridors of migrating neurons in the human brain and malignancies. Future Oncol 7: 339–354. their decline during infancy. Nature 478: 382–386. Reardon DA, Turner S, Peters KB, Desjardins A, Gururangan S, Sandsmark DK, Zhang H, Hegedus B, Pelletier CL, Weber JD, Sampson JH, McLendon RE, Herndon JE II, Jones LW, Gutmann DH. 2007. Nucleophosmin mediates mammalian Kirkpatrick JP, et al. 2011b. A review of VEGF/VEGFR-targeted target of rapamycin-dependent actin cytoskeleton dynamics therapeutics for recurrent glioblastoma. JNatlComprCanc and proliferation in neurofibromin-deficient astrocytes. Can- Netw 9: 414–427. cer Res 67: 4790–4799. Reifenberger G, Reifenberger J, Ichimura K, Meltzer PS, Collins Sanson M, Marie Y, Paris S, Idbaih A, Laffaire J, Ducray F, El VP. 1994. Amplification of multiple genes from chromo- Hallani S, Boisselier B, Mokhtari K, Hoang-Xuan K, et al. somal region 12q13-14 in human malignant gliomas: Pre- 2009. Isocitrate dehydrogenase 1 codon 132 mutation is an liminary mapping of the amplicons shows preferential in- important prognostic biomarker in gliomas. J Clin Oncol 27: volvement of CDK4, SAS, and MDM2. Cancer Res 54: 4299– 4150–4154. 4303. Sarbassov DD, Guertin DA, Ali SM, Sabatini DM. 2005. Reitman ZJ, Jin G, Karoly ED, Spasojevic I, Yang J, Kinzler KW, Phosphorylation and regulation of Akt/PKB by the rictor– He Y, Bigner DD, Vogelstein B, Yan H. 2011. Profiling the mTOR complex. Science 307: 1098–1101.

GENES & DEVELOPMENT 781 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al.

Sathornsumetee S, Rich JN. 2007. Antiangiogenic therapy in drogliomas with grade IV anaplastic features. J Neuropathol malignant glioma: Promise and challenge. Curr Pharm Des Exp Neurol 59: 495–503. 13: 3545–3558. Snuderl M, Eichler AF, Ligon KL, Vu QU, Silver M, Betensky Scherer H. 1940. The forms of growth in gliomas and their RA, Ligon AH, Wen PY, Louis DN, Iafrate AJ. 2009. Poly- practical signficance. Brain 63: 1–35. somy for chromosomes 1 and 19 predicts earlier recurrence Schindler G, Capper D, Meyer J, Janzarik W, Omran H, Herold- in anaplastic oligodendrogliomas with concurrent 1p/19q Mende C, Schmieder K, Wesseling P, Mawrin C, Hasselblatt loss. Clin Cancer Res 15: 6430–6437. M, et al. 2011. Analysis of BRAF V600E mutation in 1,320 Snuderl M, Fazlollahi L, Le LP, Nitta M, Zhelyazkova BH, nervous system tumors reveals high mutation frequencies in Davidson CJ, Akhavanfard S, Cahill DP, Aldape KD, Betensky pleomorphic xanthoastrocytoma, ganglioglioma and extra- RA, et al. 2011. Mosaic amplification of multiple recep- cerebellar pilocytic astrocytoma. Acta Neuropathol 121: tor tyrosine kinase genes in glioblastoma. Cancer Cell 20: 397–405. 810–817. Schlegel J, Stumm G, Brandle K, Merdes A, Mechtersheimer G, Sohr S, Engeland K. 2008. RHAMM is differentially expressed in Hynes NE, Kiessling M. 1994. Amplification and differential the cell cycle and downregulated by the tumor suppressor expression of members of the erbB-gene family in human p53. Cell Cycle 7: 3448–3460. glioblastoma. J Neurooncol 22: 201–207. Song H, Li Y, Lee J, Schwartz AL, Bu G. 2009. Low-density Schmidt EE, Ichimura K, Reifenberger G, Collins VP. 1994. lipoprotein receptor-related protein 1 promotes cancer cell CDKN2 (/MTS1) gene deletion or CDK4 amplification migration and invasion by inducing the expression of matrix occurs in the majority of glioblastomas. Cancer Res 54: metalloproteinases 2 and 9. Cancer Res 69: 879–886. 6321–6324. Song MS, Carracedo A, Salmena L, Song SJ, Egia A, Malumbres Schuller U, Heine VM, Mao J, Kho AT, Dillon AK, Han YG, M, Pandolfi PP. 2011. Nuclear PTEN regulates the APC– Huillard E, Sun T, Ligon AH, Qian Y, et al. 2008. Acquisition CDH1 tumor-suppressive complex in a phosphatase-inde- of granule neuron precursor identity is a critical determinant pendent manner. Cell 144: 187–199. of progenitor cell competence to form Shh-induced medul- Soroceanu L, Kharbanda S, Chen R, Soriano RH, Aldape K, loblastoma. Cancer Cell 14: 123–134. Misra A, Zha J, Forrest WF, Nigro JM, Modrusan Z, et al. Sequist LV, Heist RS, Shaw AT, Fidias P, Rosovsky R, Temel JS, 2007. Identification of IGF2 signaling through phosphoinosi- Lennes IT, Digumarthy S, Waltman BA, Bast E, et al. 2011. tide-3-kinase regulatory subunit 3 as a growth-promoting Implementing multiplexed genotyping of non-small-cell axis in glioblastoma. Proc Natl Acad Sci 104: 3466–3471. lung cancers into routine clinical practice. Ann Oncol 22: Stettner MR, Wang W, Nabors LB, Bharara S, Flynn DC, 2616–2624. Grammer JR, Gillespie GY, Gladson CL. 2005. Lyn kinase Sharma SV, Settleman J. 2007. Oncogene addiction: Setting the activity is the predominant cellular SRC kinase activity in stage for molecularly targeted cancer therapy. Genes Dev 21: glioblastoma tumor cells. Cancer Res 65: 5535–5543. 3214–3231. Stockhausen MT, Broholm H, Villingshoj M, Kirchhoff M, Shervington A, Lu C. 2008. Expression of multidrug resistance Gerdes T, Kristoffersen K, Kosteljanetz M, Spang-Thomsen genes in normal and cancer stem cells. Cancer Invest 26: M, Poulsen HS. 2011. Maintenance of EGFR and EGFRvIII 535–542. expressions in an in vivo and in vitro model of human Shinojima N, Tada K, Shiraishi S, Kamiryo T, Kochi M, glioblastoma multiforme. Exp Cell Res 317: 1513–1526. Nakamura H, Makino K, Saya H, Hirano H, Kuratsu J, Stommel JM, Kimmelman AC, Ying H, Nabioullin R, Ponugoti et al. 2003. Prognostic value of epidermal growth factor AH, Wiedemeyer R, Stegh AH, Bradner JE, Ligon KL, Brennan receptor in patients with glioblastoma multiforme. Cancer C, et al. 2007. Coactivation of receptor tyrosine kinases affects Res 63: 6962–6970. the response of tumor cells to targeted therapies. Science 318: Shweiki D, Itin A, Soffer D, Keshet E. 1992. Vascular endothelial 287–290. growth factor induced by hypoxia may mediate hypoxia- Stratton MR, Campbell PJ, Futreal PA. 2009. The cancer initiated angiogenesis. Nature 359: 843–845. genome. Nature 458: 719–724. Singh SK, Clarke ID, Terasaki M, Bonn VE, Hawkins C, Squire J, Stupp R, Mason WP, van den Bent MJ, Weller M, Fisher B, Dirks PB. 2003. Identification of a cancer stem cell in human Taphoorn MJ, Belanger K, Brandes AA, Marosi C, Bogdahn U, brain tumors. Cancer Res 63: 5821–5828. et al. 2005. Radiotherapy plus concomitant and adjuvant Singh SK, Hawkins C, Clarke ID, Squire JA, Bayani J, Hide T, temozolomide for glioblastoma. N Engl J Med 352: 987–996. Henkelman RM, Cusimano MD, Dirks PB. 2004. Identification Sugiarto S, Persson AI, Munoz EG, Waldhuber M, Lamagna C, of human brain tumour initiating cells. Nature 432: 396–401. Andor N, Hanecker P, Ayers-Ringler J, Phillips J, Siu J, et al. Sirko S, von Holst A, Wizenmann A, Gotz M, Faissner A. 2007. 2011. Asymmetry-defective oligodendrocyte progenitors are Chondroitin sulfate glycosaminoglycans control prolifera- glioma precursors. Cancer Cell 20: 328–340. tion, radial glia cell differentiation and neurogenesis in Sun M, Hillmann P, Hofmann BT, Hart JR, Vogt PK. 2010. neural stem/progenitor cells. Development 134: 2727–2738. Cancer-derived mutations in the regulatory subunit p85a of Sjoblom T, Jones S, Wood LD, Parsons DW, Lin J, Barber TD, phosphoinositide 3-kinase function through the catalytic Mandelker D, Leary RJ, Ptak J, Silliman N, et al. 2006. The subunit p110a. Proc Natl Acad Sci 107: 15547–15552. consensus coding sequences of human breast and colorectal Sunayama J, Sato A, Matsuda K, Tachibana K, Suzuki K, Narita cancers. Science 314: 268–274. Y, Shibui S, Sakurada K, Kayama T, Tomiyama A, et al. 2010. Slamon DJ, Leyland-Jones B, Shak S, Fuchs H, Paton V, Bajamonde Dual blocking of mTor and PI3K elicits a prodifferentiation A, Fleming T, Eiermann W, Wolter J, Pegram M, et al. 2001. effect on glioblastoma stem-like cells. Neuro-oncol 12: 1205– Use of chemotherapy plus a monoclonal antibody against 1219. HER2 for that overexpresses HER2. Szerlip NJ, Pedraza A, Chakravarty D, Azim M, McGuire J, Fang N Engl J Med 344: 783–792. Y, Ozawa T, Holland EC, Huse JT, Jhanwar S, et al. 2012. Smith JS, Wang XY, Qian J, Hosek SM, Scheithauer BW, Jenkins Intratumoral heterogeneity of EGFR and PDGFRA amplifi- RB, James CD. 2000. Amplification of the platelet-derived cation in glioblastoma defines subpopulations with distinct growth factor receptor-A (PDGFRA) gene occurs in oligoden- growth factor response. Proc Natl Acad Sci 109: 3041–3046.

782 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Molecular and cellular basis of glioblastoma

Tate MC, Aghi MK. 2009. Biology of angiogenesis and invasion recurrent glioblastoma multiforme. J Clin Oncol 25: 4722– in glioma. Neurotherapeutics 6: 447–457. 4729. Taylor CM, Sauvageot CM, Wen PYC, Stiles CD. 2010. Cellular Wakimoto H, Mohapatra G, Kanai R, Curry WT Jr, Yip S, Nitta origins of malignant glioma: The cancer stem cell polemic. M, Patel AP, Barnard ZR, Stemmer-Rachamimov AO, Louis In Primary central nervous system tumors: Pathogenesis DN, et al. 2012. Maintenance of primary tumor phenotype and therapy (ed. AD Norden et al.), pp. 45–54. Springer, New and genotype in glioblastoma stem cells. Neuro-oncol 14: York. 132–144. Teodorczyk M, Martin-Villalba A. 2010. Sensing invasion: Cell Wang SC, Hung MC. 2009. Nuclear translocation of the epider- surface receptors driving spreading of glioblastoma. J Cell mal growth factor receptor family membrane tyrosine kinase Physiol 222: 1–10. receptors. Clin Cancer Res 15: 6484–6489. Tian Y, Rich BE, Vena N, Craig JM, Macconaill LE, Rajaram V, Wang H, Shen W, Huang H, Hu L, Ramdas L, Zhou YH, Liao WS, Goldman S, Taha H, Mahmoud M, Ozek M, et al. 2011. Fuller GN, Zhang W. 2003. Insulin-like growth factor bind- Detection of KIAA1549–BRAF fusion transcripts in formalin- ing protein 2 enhances glioblastoma invasion by activating fixed paraffin-embedded pediatric low-grade gliomas. J Mol invasion-enhancing genes. Cancer Res 63: 4315–4321. Diagn 13: 669–677. Wang H, Zhang W, Huang HJ, Liao WS, Fuller GN. 2004. Tso CL, Freije WA, Day A, Chen Z, Merriman B, Perlina A, Lee Analysis of the activation status of Akt, NFkB, and Stat3 in Y, Dia EQ, Yoshimoto K, Mischel PS, et al. 2006. Distinct human diffuse gliomas. Lab Invest 84: 941–951. transcription profiles of primary and secondary glioblastoma Wang X, Trotman LC, Koppie T, Alimonti A, Chen Z, Gao Z, subgroups. Cancer Res 66: 159–167. Wang J, Erdjument-Bromage H, Tempst P, Cordon-Cardo C, Turcan S, Rohle D, Goenka A, Walsh LA, Fang F, Yilmaz E, et al. 2007. NEDD4-1 is a proto-oncogenic ubiquitin ligase Campos C, Fabius AWM, Lu C, Ward PS, et al. 2012. IDH1 for PTEN. Cell 128: 129–139. mutation is sufficient to establish the glioma hypermethy- Wang Y, Yang J, Zheng H, Tomasek GJ, Zhang P, McKeever PE, lator phenotype. Nature doi: 10.1038/nature10866. Lee EY, Zhu Y. 2009. Expression of mutant p53 proteins Ueki K, Ono Y, Henson JW, Efird JT, von Deimling A, Louis DN. implicates a lineage relationship between neural stem cells 1996. CDKN2/p16 or RB alterations occur in the majority of and malignant astrocytic glioma in a murine model. Cancer glioblastomas and are inversely correlated. Cancer Res 56: Cell 15: 514–526. 150–153. Wang R, Chadalavada K, Wilshire J, Kowalik U, Hovinga KE, Vasudevan KM, Barbie DA, Davies MA, Rabinovsky R, McNear Geber A, Fligelman B, Leversha M, Brennan C, Tabar V. 2010. CJ, Kim JJ, Hennessy BT, Tseng H, Pochanard P, Kim SY, Glioblastoma stem-like cells give rise to tumour endothe- et al. 2009. AKT-independent signaling downstream of lium. Nature 468: 829–833. oncogenic PIK3CA mutations in human cancer. Cancer Cell Ward PS, Patel J, Wise DR, Abdel-Wahab O, Bennett BD, Coller 16: 21–32. HA, Cross JR, Fantin VR, Hedvat CV, Perl AE, et al. 2010. Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson The common feature of leukemia-associated IDH1 and IDH2 MD, Miller CR, Ding L, Golub T, Mesirov JP, et al. 2010. mutations is a neomorphic enzyme activity converting Integrated genomic analysis identifies clinically relevant a-ketoglutarate to 2-hydroxyglutarate. Cancer Cell 17: 225– subtypes of glioblastoma characterized by abnormalities in 234. PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 17: 98–110. Warrington NM, Gianino SM, Jackson E, Goldhoff P, Garbow JR, Verhoeff JJ, van Tellingen O, Claes A, Stalpers LJ, van Linde ME, Piwnica-Worms D, Gutmann DH, Rubin JB. 2010. Cyclic Richel DJ, Leenders WP, van Furth WR. 2009. Concerns AMP suppression is sufficient to induce gliomagenesis in about anti-angiogenic treatment in patients with glioblas- a mouse model of neurofibromatosis-1. Cancer Res 70: 5717– toma multiforme. BMC Cancer 9: 444. doi: 10.1186/1471- 5727. 2407-9-444. Watnick RS, Cheng YN, Rangarajan A, Ince TA, Weinberg RA. Visvader JE. 2011. Cells of origin in cancer. Nature 469: 314– 2003. Ras modulates Myc activity to repress thrombospon- 322. din-1 expression and increase tumor angiogenesis. Cancer Vivanco I, Rohle D, Versele M, Iwanami A, Kuga D, Oldrini B, Cell 3: 219–231. Tanaka K, Dang J, Kubek S, Palaskas N, et al. 2010. The Wee B, Charles N, Holland EC. 2012. Animal models to study phosphatase and tensin homolog regulates epidermal growth cancer-initiating cells from glioblastoma. Front Biosci 17: factor receptor (EGFR) inhibitor response by targeting EGFR 2243–2258. for degradation. Proc Natl Acad Sci 107: 6459–6464. Weir B, Zhao X, Meyerson M. 2004. Somatic alterations in the Vogelstein B, Kinzler KW. 2004. Cancer genes and the pathways human cancer genome. Cancer Cell 6: 433–438. they control. Nat Med 10: 789–799. Weis SM, Cheresh DA. 2011. Tumor angiogenesis: Molecular Vogelstein B, Fearon ER, Hamilton SR, Kern SE, Preisinger AC, pathways and therapeutic targets. Nat Med 17: 1359–1370. Leppert M, Nakamura Y, White R, Smits AM, Bos JL. 1988. Wen PY, Macdonald DR, Reardon DA, Cloughesy TF, Sorensen Genetic alterations during colorectal-tumor development. N AG, Galanis E, Degroot J, Wick W, Gilbert MR, Lassman AB, Engl J Med 319: 525–532. et al. 2010. Updated response assessment criteria for high- von Deimling A, Korshunov A, Hartmann C. 2011. The next grade gliomas: Response assessment in neuro-oncology generation of glioma biomarkers: MGMT methylation, working group. J Clin Oncol 28: 1963–1972. BRAF fusions and IDH1 mutations. Brain Pathol 21: 74–87. Whiteman DC, Zhou XP, Cummings MC, Pavey S, Hayward Vredenburgh JJ, Desjardins A, Herndon JE II, Dowell JM, Reardon NK, Eng C. 2002. Nuclear PTEN expression and clinicopath- DA, Quinn JA, Rich JN, Sathornsumetee S, Gururangan S, ologic features in a population-based series of primary Wagner M, et al. 2007a. Phase II trial of bevacizumab and cutaneous melanoma. Int J Cancer 99: 63–67. irinotecan in recurrent malignant glioma. Clin Cancer Res Wiedemeyer R, Brennan C, Heffernan TP, Xiao Y, Mahoney J, 13: 1253–1259. Protopopov A, Zheng H, Bignell G, Furnari F, Cavenee WK, Vredenburgh JJ, Desjardins A, Herndon JE II, Marcello J, Reardon et al. 2008. Feedback circuit among INK4 tumor suppressors DA, Quinn JA, Rich JN, Sathornsumetee S, Gururangan S, constrains human glioblastoma development. Cancer Cell Sampson J, et al. 2007b. Bevacizumab plus irinotecan in 13: 355–364.

GENES & DEVELOPMENT 783 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Dunn et al.

Wiedemeyer WR, Dunn IF, Quayle SN, Zhang J, Chheda MG, Ziegler DS, Wright RD, Kesari S, Lemieux ME, Tran MA, Jain Dunn GP, Zhuang L, Rosenbluh J, Chen S, Xiao Y, et al. 2010. M, Zawel L, Kung AL. 2008. Resistance of human glioblas- Pattern of retinoblastoma pathway inactivation dictates re- toma multiforme cells to growth factor inhibitors is over- sponse to CDK4/6 inhibition in GBM. Proc Natl Acad Sci come by blockade of inhibitor of apoptosis proteins. J Clin 107: 11501–11506. Invest 118: 3109–3122. Wong K, Young GS, Makale M, Hu X, Yildirim N, Cui K, Wong Zong H, Espinosa JS, Su HH, Muzumdar MD, Luo L. 2005. Mosaic ST, Kesari S. 2011. Characterization of a human tumorsphere analysis with double markers in mice. Cell 121: 479–492. glioma orthotopic model using magnetic resonance imaging. J Neurooncol 104: 473–481. Wu G, Broniscer A, McEachron TA, Lu C, Paugh BS, Becksfort J, Qu C, Ding L, Huether R, Parker M, et al. 2012. Somatic histone H3 alterations in pediatric diffuse intrinsic pontine gliomas and non-brainstem glioblastomas. Nat Genet 44: 251–253. Wykosky J, Gibo DM, Stanton C, Debinski W. 2005. EphA2 as a novel molecular marker and target in glioblastoma multi- forme. Mol Cancer Res 3: 541–551. Xu W, Yang H, Liu Y, Yang Y, Wang P, Kim SH, Ito S, Yang C, Xiao MT, Liu LX, et al. 2011. Oncometabolite 2-hydroxyglu- tarate is a competitive inhibitor of a-ketoglutarate-depen- dent dioxygenases. Cancer Cell 19: 17–30. Yan H, Parsons DW, Jin G, McLendon R, Rasheed BA, Yuan W, Kos I, Batinic-Haberle I, Jones S, Riggins GJ, et al. 2009. IDH1 and IDH2 mutations in gliomas. N Engl J Med 360: 765–773. Yao Z, Fenoglio S, Gao DC, Camiolo M, Stiles B, Lindsted T, Schlederer M, Johns C, Altorki N, Mittal V, et al. 2010. TGF- b IL-6 axis mediates selective and adaptive mechanisms of resistance to molecular targeted therapy in lung cancer. Proc Natl Acad Sci 107: 15535–15540. Yip S, Butterfield YS, Morozova O, Chittaranjan S, Blough MD, An J, Birol I, Chesnelong C, Chiu R, Chuah E, et al. 2012. Concurrent CIC mutations, IDH mutations, and 1p/19q loss distinguish oligodendrogliomas from other cancers. J Pathol 226: 7–16. Yordanova YN, Moritz-Gasser S, Duffau H. 2011. Awake surgery for WHO Grade II gliomas within ‘noneloquent’ areas in the left dominant hemisphere: Toward a ‘supratotal’ resection. Clinical article. J Neurosurg 115: 232–239. Yuan TL, Cantley LC. 2008. PI3K pathway alterations in cancer: Variations on a theme. Oncogene 27: 5497–5510. Zhao S, Lin Y, Xu W, Jiang W, Zha Z, Wang P, Yu W, Li Z, Gong L, Peng Y, et al. 2009. Glioma-derived mutations in IDH1 dominantly inhibit IDH1 catalytic activity and induce HIF- 1a. Science 324: 261–265. Zheng H, Ying H, Yan H, Kimmelman AC, Hiller DJ, Chen AJ, Perry SR, Tonon G, Chu GC, Ding Z, et al. 2008. p53 and Pten control neural and glioma stem/progenitor cell renewal and differentiation. Nature 455: 1129–1133. Zheng H, Ying H, Wiedemeyer R, Yan H, Quayle SN, Ivanova EV, Paik JH, Zhang H, Xiao Y, Perry SR, et al. 2010. PLAGL2 regulates Wnt signaling to impede differentiation in neural stem cells and gliomas. Cancer Cell 17: 497–509. Zhou XP, Loukola A, Salovaara R, Nystrom-Lahti M, Peltomaki P, de la Chapelle A, Aaltonen LA, Eng C. 2002. PTEN mutational spectra, expression levels, and subcellular local- ization in microsatellite stable and unstable colorectal can- cers. Am J Pathol 161: 439–447. Zhu Y, Guignard F, Zhao D, Liu L, Burns DK, Mason RP, Messing A, Parada LF. 2005. Early inactivation of p53 tumor suppressor gene cooperating with NF1 loss induces malig- nant astrocytoma. Cancer Cell 8: 119–130. Zhu H, Acquaviva J, Ramachandran P, Boskovitz A, Woolfenden S, Pfannl R, Bronson RT, Chen JW, Weissleder R, Housman DE, et al. 2009. Oncogenic EGFR signaling cooperates with loss of tumor suppressor gene functions in gliomagenesis. Proc Natl Acad Sci 106: 2712–2716.

784 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press

Emerging insights into the molecular and cellular basis of glioblastoma

Gavin P. Dunn, Mikael L. Rinne, Jill Wykosky, et al.

Genes Dev. 2012, 26: Access the most recent version at doi:10.1101/gad.187922.112

References This article cites 349 articles, 114 of which can be accessed free at: http://genesdev.cshlp.org/content/26/8/756.full.html#ref-list-1

License

Email Alerting Receive free email alerts when new articles cite this article - sign up in the box at the top Service right corner of the article or click here.

Copyright © 2012 by Cold Spring Harbor Laboratory Press